JP5351925B2 - Inspection device and inspection method for long member for transfer mechanism including steel cord - Google Patents

Abstract

The invention provides a testing device and testing method for a strip component for a transportation mechanism of a steel wire. According to the state of the steel wire of the strip component for a transportation mechanism, such as the steel wire embedded in a handrail of a passenger conveyer, the quality of the handril is evaluated. The testing device comprises an X ray shooting unit and an image processing unit, wherein the X ray shooting unit uses the X ray to shoot the handrail of the passenger conveyer; the image processing unit process the image shoot by the X ray shooting unit and detects the steel wire which has monofilament situation for the embedded steel wire; and when the portion for having the monofilament situation on the steel wire along the longitudinal directon of the handrail achieves the prescribed length continuously, the quality of the handrail is degraded.

Description

  TECHNICAL FIELD The present invention relates to an inspection device and inspection method for long members used in a transfer mechanism, such as escalators, handrails of passenger conveyors such as moving walkways, wire ropes for elevators, etc., in particular, long members for transfer mechanisms including steel cords. It is.

  Passenger conveyors such as escalators and moving walkways are provided with handrails that are handrails that move in synchronism with the steps on which the passengers are mounted. This handrail has a structure in which a steel wire (or steel belt) called a plurality of steel cords (hereinafter may be abbreviated as SC) is fixed with a holding rubber and the outside is covered with a skin of a cosmetic rubber. Yes.

  The steel cord is formed by twisting several thin wires called strands into a slightly thick wire called a strand, and twisting these several wires. When such a steel cord deteriorates over time, the strands are unwound and the strands are recognized in a state of jumping out of the steel cord. Hereinafter, a state in which the wire jumps out of the steel cord through such a process is referred to as “strengthening (or fraying)” of the steel cord. At this stage, however, the wire often does not jump out of the outer skin of the decorative rubber.

  During operation of the passenger conveyor, the steel cord built into the handrail is repeatedly bent by the action of the driving device. Eventually, “stranding” occurs. It can be diagnosed whether or not the deterioration of the steel cord has started by the presence or absence of this “stranding”. In addition, it is possible to diagnose the degree of progress of deterioration in accordance with the length of the state in which this “stranding” has occurred in the longitudinal direction of the handrail.

  In Patent Document 1, light is projected onto a crane rope to be diagnosed, and the transmitted light is received by a light receiver to convert the amount of received light into a signal. It is disclosed to determine (cut, wear, deformation, elongation).

  Patent Document 2 discloses an apparatus and a method for detecting a wear leg generated on a strand from a rope image photographed by a camera and determining deterioration of the wire rope based on the number of the wear legs as a strand break detection means. Has been.

  In Patent Literature 3, the outer diameter of the wire rope was measured by measuring the outer diameter of the wire rope from the output signal of the light receiver that received the laser projected on the wire rope, with the laser projector and the receiver sandwiching the wire rope. It is disclosed that the deterioration is judged from the disturbance of the periodicity of the outer diameter change.

  In addition, as a means of diagnosing the deterioration of the steel cord of the passenger conveyor handrail, in Patent Document 4, X-ray transmission imaging is performed with a handrail incorporating a steel cord interposed, and the imaging is performed by moving as necessary. It is disclosed that the state of the steel cord is visually confirmed.

  In Patent Document 5, a power spectrum, which is the square of a two-dimensional Fourier spectrum, is calculated from an X-ray image taken with a configuration similar to that of Patent Document 4, and the intervals of a plurality of steel cords lined up from this power spectrum pattern are uniform. It is disclosed to automatically determine the presence or absence of sex and meandering.

  Although it does not diagnose the deterioration of the steel cord of the passenger conveyor handrail, in Patent Document 6, the steel belt in the rubber tire is X-rayed and the images continuously taken in the tire circumferential direction are pasted to create a panorama. After that, image processing is performed to extract the contour of the steel belt, and the width in the direction orthogonal to the circumference is measured at regular intervals in the circumferential direction. If the width is discontinuous, it is determined as defective. It is disclosed.

  In addition to the above-mentioned patent documents, the following document presentation list presents non-patent documents 1 and 2, but this non-patent document is a space in a part of image processing in an embodiment of the present invention to be described later. Since it is used for filtering, it will be described in the section of the embodiment.

JP-A-11-325844 JP 2009-12903 A JP 2008-214037 A Japanese Patent Laid-Open No. 10-10060 JP 2005-126175 A JP 2008-309649 A

Digital image processing, published by CG-ARTS Association (2006), p. 106-110. Digital image processing, published by CG-ARTS Association (2006), p. 114-121.

  It is necessary to accurately diagnose the stage of deterioration in order to appropriately grasp the replacement time of the passenger conveyor handrail. Passenger conveyor handrails gradually deteriorate over time, resulting in "stranding" in the built-in steel cord. Therefore, if it becomes possible to automatically detect the presence / absence of “stranding” and the length of “stranding” in the longitudinal direction of the handrail, it is possible to automate deterioration diagnosis. Become.

  In the technique disclosed in Patent Document 1, the amount of received light projected onto the wire rope of a crane or elevator is a signal of the amount of received light because the fluctuation of the rope silhouette width based on the twist of the normal wire rope is periodic. If it deviates from a periodic change (in this case, an unspecified frequency spectrum occurs), it is determined as degradation. Since the ropes of cranes and elevators are premised on continuous, there may be cases where it is acceptable to deviate from periodic changes. On the other hand, the handrail of the passenger conveyor has deviated from the periodic change because the joint of the built-in steel cord is non-degraded and has a gap and is aperiodic. Such a deterioration judgment method may be inappropriate.

  The technique disclosed in Patent Document 2 is to detect the number of worn legs of a wire from each frame of an ITV image captured by an industrial television and detect the number of worn wires from the number of wear. No specific image processing for detection is disclosed.

  The technique disclosed in Patent Document 3 measures changes in the outer diameter of the wire rope by projecting laser light. As in the case of Patent Document 1, this is not suitable for diagnosis of a handrail in which a non-deteriorated product has a gap in a steel cord. In addition, no consideration is given to detecting “stranding” that is important for deterioration determination.

  Since the technique disclosed in Patent Document 4 determines the deterioration of the wire rope or the steel cord by visual observation of maintenance personnel, variations in determination are likely to occur due to individual differences among maintenance personnel.

  The technique disclosed in Patent Document 5 quantitatively determines the presence or absence of steel cord non-uniformity or meandering based on the power spectrum. Such inhomogeneity and skew of the built-in steel cord are also observed in the non-degraded handrail, and therefore, it is difficult to say that the present technology judges whether or not the steel cord itself has deteriorated. In addition, no consideration is given to a technique for detecting “stranding” that is important for deterioration determination.

  The technique disclosed in Patent Document 6 determines that the measured value of the width of the steel belt deviates from a predetermined range on the assumption that the width of the steel belt of a normal tire is constant. On the other hand, steel cords built in handrails usually meander or fluctuate in the direction perpendicular to the longitudinal direction, and it is not possible to determine whether or not steel cords have deteriorated by this technology. . Also in this technology, there is no disclosure of a technology for detecting “stranding” required for deterioration determination.

  As described above, several known techniques that the applicant can know have been introduced. However, a technique for automatically diagnosing the deterioration (break of wire, wear, deformation, elongation) of a steel cord such as a wire rope is disclosed in Patent Document 1. Although it is proposed as shown in Fig. 1, it is not suitable for the deterioration diagnosis of steel cords for passenger conveyor handrails as described above. No technology has been proposed.

  The present invention has been made in view of the above circumstances, and the object of the present invention is to “strand” long members including steel cords of handrails such as passenger conveyors and other transfer mechanisms. It is an object of the present invention to provide an inspection device for a long member for a transfer mechanism that can be specialized and automatically detected and / or diagnosed with high accuracy.

In order to achieve the above object, the present invention basically includes:
An inspection device for a long member including a steel cord used for a transfer mechanism,
i) an imaging unit that forms a projection image of a steel cord by irradiating X-rays or visible light;
ii) An image processing for inputting the projection image and extracting the line segment when the input projection image includes a line segment that is thinner than a normal steel code and higher in luminance than a normal steel code. And an image processing unit for detecting “stirring” (fraying) of the steel cord to be inspected based on the line segment extraction data.

  Automatic detection and / or diagnosis can be performed with high accuracy by specializing in “stranding” long members including steel cords of handrails such as passenger conveyors and other transfer mechanisms.

  Other effects of the invention according to the dependent claims of the present application will be made clear from the description of the entire specification.

The figure which shows the strand detection processing flow of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the example of the line segment detection filter of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the example of the line segment total image of the handrail inspection apparatus of the passenger conveyor of this invention. The schematic diagram which shows the equipment structure of the handrail inspection apparatus of the passenger conveyor of this invention, and the example of mounting to a handrail. FIG. 5 is a schematic diagram showing a device configuration of a handrail inspection device for a passenger conveyor according to the present invention and an example of mounting on the handrail in a manner different from FIG. The figure which shows the example of the handrail X-ray image image | photographed with the handrail inspection apparatus of the passenger conveyor of this invention. The conceptual diagram of the good quality handrail X-ray image of the handrail inspection apparatus of the passenger conveyor of this invention. The conceptual diagram of the deterioration initial handrail X-ray image of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the example of an imaging | photography system. The figure explaining the luminance distribution of the image which a normal steel cord forms. The figure explaining the luminance distribution of the image which the thin steel cord made into a strand forms. The figure explaining the luminance distribution of the image which the steel cord which has a boundary outer diameter forms. The figure explaining the luminance distribution of the image which the steel cord which has th outer diameter forms. The figure which shows the example of an imaging | photography system. The figure which shows one Example of the image process part of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the process 1st example of the SC detection means of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows an example of the contrast correction | amendment of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows an example of the detection of the steel cord candidate cell of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the connection process of the SC candidate cell of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the process 2nd example of the SC detection means of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows an example of the collation method of the SC model of the handrail inspection apparatus of the passenger conveyor of this invention, and the detected steel code. The figure which shows an example of the collation method of the SC model of the handrail inspection apparatus of the passenger conveyor of this invention, and the detected steel code. The figure which shows the distance table used for an example of the collation method of the SC model of the handrail inspection apparatus of the passenger conveyor of this invention, and the detected steel code. The figure which shows the processing flow of the SC model holding | maintenance means and SC trace / element feature detection means of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the detection method of the element characteristic "missing" of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the detection method of the element characteristic "contact" of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the detection method of the element characteristic "entanglement" of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the example of stranding of the handrail inspection apparatus of the passenger conveyor of this invention. The figure which shows the example of the quality determination conditions of the handrail inspection apparatus of the passenger conveyor of this invention. The figure explaining the brightness | luminance measuring method when the brightness | luminance threshold value Bth is non-fixed. The figure which shows the 2nd Example of the image process part of the handrail inspection apparatus of a passenger conveyor. The figure which shows the example of SC detection log table and SC model log table. The figure which shows another example of SC detection log table and SC model log table. The figure which shows the method of implement | achieving a bidirectional trace. The figure which shows the flow of a process of the image processing means of the handrail inspection apparatus of the passenger conveyor of this invention in the case of performing a bidirectional | two-way trace. The figure which shows the flow of a process of the image process means of the handrail inspection apparatus of the passenger conveyor of this invention. The figure of the imaging | photography system for detecting a strand with visible light.

  In the present invention, a long member including a steel cord used for a transfer mechanism is to be inspected. First, as a long member of a transfer mechanism, a handrail of a passenger conveyor such as an escalator or a moving sidewalk is a preferred example. . Moreover, the bare wire rope used for an elevator may be sufficient as a long member of a transfer mechanism.

  The present invention includes an imaging unit that forms a projection image of a steel cord by irradiating X-rays or visible light. When the steel cord of a passenger conveyor handrail is to be imaged, the steel cord is a decorative rubber. An imaging unit using X-rays is preferable because it is covered with the outer skin. Moreover, when taking a bare wire rope (steel cord) as an imaging object like an elevator, an imaging unit using visible light is preferable.

For example, when the image processing unit inputs a projection image from the photographing unit and the input projection image includes a line segment that is thinner than a normal steel code and brighter than a normal steel code, Image processing is performed to extract the line segment as a “strand” candidate. As a preferred example, for example, the photographing unit is set to form a main shadow for a steel cord having a normal outer diameter and to form a thin penumbra for a frayed steel cord. Then, the image processing unit sets a luminance range that is a determination criterion for penumbra and extracts a line segment corresponding to fraying based on the determination criterion.
[Visualization and imaging of X-ray images]
In the present embodiment, an X-ray image of the passenger conveyor handrail is processed to detect the “stranding” of the steel cord and determine the deterioration. At that time, the “crimping” of the steel cord is based on whether or not those features continue for a predetermined length or more in the longitudinal direction of the handrail. Then, using the detected features, the quality of the handrail is evaluated, for example, in two stages of deterioration and non-defective product or in three or more stages of defect, deterioration, and non-defective product.

In the present embodiment, the steel cord built in the handrail of the passenger conveyor is visualized by X-ray imaging and image processing is performed. X-ray imaging uses a steel cord built into the handrail between an X-ray tube that emits X-rays radially and a scintillator that is a screen for visualizing X-ray images in an environment where all external light is blocked. This is done with the arrangement. The scintillator is a paper or resin having a thickness of about several hundreds μm coated with a fluorescent material and emits light while being irradiated with X-rays. Of these, the portion irradiated with a large X-ray dose as it is without hitting the built-in steel cord emits light with high brightness. On the other hand, the portion that is absorbed or reflected by the steel cord and is not irradiated with X-rays does not emit light. However, there are actually X-rays that are diffusely reflected even at the part blocked by the steel cord, and emit a little light with low brightness. Since a relative luminance distribution is generated as a picture of the steel cord, a configuration is adopted in which this is photographed with a camera.
[Description of strands]
Here, “stranding” is a state in which a thin wire called a strand constituting a steel cord is unwound and exists alone. The outer diameter of the strand is about 1/5 to 1/10 of the outer diameter of a normal steel cord. For example, if the outer diameter of the strand is about 0.18 mm, the outer diameter of the steel cord formed by combining these is about 1.5 mm to about 1.8 mm.
[Realization of strands]
The physical index that distinguishes ordinary steel cords and strands is their outer diameter. The outer diameter of ordinary steel cords is 1.5 to 1.8 mm, while the outer diameter of strands is 0.15 to 1.8 mm. The range is 0.3 mm. However, even when shooting with an optical lens system in the visible light region, it is difficult to obtain a clear image before the outer diameter can be measured due to fluctuations in the distance between the camera and the strand. In addition, when taking an X-ray image of an element wire in a steel cord built in a handrail, it is necessary to irradiate a strong X-ray dose through a very small opening of less than 0.1 mm in order to clearly capture an extremely thin element wire. There is. However, the output of X-rays is limited due to cooling, safety, and cost restrictions, and the aperture size for X-ray emission needs to be about 0.7 mm. Then, it is necessary to capture and automatically detect a strand having an outer diameter smaller than that while photographing with an X-ray exit aperture having a certain size.

As will be described later with reference to the drawings, when a steel cord is photographed with X-rays emitted from a large aperture, a main shadow, which is a dark shadow formed by completely shielding the X-rays with the steel cord, is generated. And since the X-ray exit opening has a size around the main shadow, a penumbra formed without partially shielding the X-ray is generated. The penumbra is a brighter shadow than the main shadow. Since the strand has an outer diameter smaller than the size of the X-ray exit aperture, a main shadow cannot be created, and only a penumbra image is obtained. Therefore, unlike ordinary steel cords with a main shadow, by detecting an object consisting only of a penumbra as an elemental wire, even an X-ray imaging system having an X-ray exit aperture larger than the elemental wire can detect the elemental wire. Is possible. Detection of thin lines by this method is not limited to X-ray imaging systems, but by detecting an object that forms only a penumbra in an image of an object projected from a light source of a finite size in a radial pattern as a strand, It is also possible to detect strands of elevator ropes and crane wires.
[Configuration example of inspection equipment]
In the present embodiment, as an example of an inspection device for a long member for a transfer mechanism, a configuration of a handrail inspection device for a passenger conveyor is exemplified, and for example, the following can be used.
(1) X-ray radiographing unit that captures X-rays of passenger conveyor handrails, and processing of images taken by the X-ray radiographing unit to “wire the steel cords built into the handrails” If the length of the portion where the “steel cord” of the steel cord is generated in the longitudinal direction of the handrail continues for a predetermined length or more, the quality of the handrail is determined to be defective. An image processing unit.
(2) In (1), the image processing unit detects “stranding” of the steel cord, and the length of the portion where the “stranding” of the steel cord occurs in the longitudinal direction of the handrail is When the length of the portion of the steel cord that is “stranded” in the longitudinal direction of the handrail is shorter than the predetermined length and is within another predetermined length range, Judge the quality as degraded.
(3) In (2), when the length of the “stranding” occurrence portion of the steel cord in the longitudinal direction of the handrail is shorter than the range of the other predetermined length, In this case, the quality of the handrail is determined as a non-defective product.

  In the passenger conveyor maintenance method according to the present embodiment, for example, the quality of the handrail is determined by a method similar to the determination method in the image processing unit in (1) to (3). Then, when it is determined that the quality of the handrail is poor, the handrail is repaired or replaced, or replaced after the repair. When it is determined that the quality of the handrail is deteriorated, the handrail determined to be deteriorated is reinspected at a cycle shorter than a normal inspection cycle.

The above-described configuration is merely an example, and can be changed as appropriate without departing from the technical idea.
[Description of Examples]
Embodiments of the present invention will be described below with reference to the drawings. In each drawing and each embodiment, the same or similar components are denoted by the same reference numerals, and redundant description is omitted.

The first embodiment exemplifies a passenger rail handrail inspection device capable of automatically diagnosing a deterioration stage of a handrail as an example of an inspection device for a long member for a transfer mechanism.
[Example of shooting configuration and wearing]
FIG. 4A is a diagram showing an apparatus configuration of a passenger rail handrail inspection device according to the present embodiment and an example in which this inspection device is mounted on a handrail at the time of inspection. It consists of part 2. The image processing unit 2 can be configured by a personal computer, for example. Further, an encoder 3 is added as necessary.

  The X-ray imaging unit 1 includes an X-ray tube 5, a scintillator 6, and a camera 7. The X-ray imaging unit 1 is configured such that the handrail 4 is positioned between the X-ray tube 5 and the scintillator 6 so that a part of the handrail 4 can be imaged and the captured image can be taken into the image processing unit. The X-rays are emitted with a spread from the X-ray tube 5 and pass through the handrail 4 to irradiate the scintillator 6. The scintillator 6 is a fluorescent plate that fluoresces when irradiated with X-rays, and fluoresces with a luminance corresponding to the amount of X-ray irradiation. The handrail 4 includes a steel cord made of steel, and a shadow image corresponding to the X-ray transmittance is generated in the scintillator 6. An image generated in the scintillator 6 is captured by the camera 7. Since the X-ray imaging unit 1 blocks light outside the apparatus, it can efficiently capture only the image generated by the light emitted from the scintillator 6. The camera 7 is connected to the image processing unit 2 and the image is taken into the image processing unit 2 as electronic data.

  Moreover, by providing the lead glass 8 between the scintillator 6 and the camera 7, it is possible to prevent the X-ray emitted from the X-ray tube 5 from damaging the camera 7. The lead glass 8 absorbs an electromagnetic wave having an X-ray wavelength and transmits an electromagnetic wave in a wavelength band of visible light related to light emission of the scintillator 6 and photographing by the camera 7. The lead glass 8 does not allow X-rays to pass through the camera 7 and allows only the light emitted by the scintillator 6 to reach the camera 7.

  FIG. 4B is a diagram showing an example in which the X-ray imaging unit is mounted on the handrail. The X-ray imaging unit 1 is mounted so that a part of the handrail 4 is taken inside. For mounting, the X-ray imaging unit 1 is a mechanism that can be separated into, for example, a camera 7, a lead rubber 8, a first part 1 a including a scintillator 6, and a second part 1 b including an X-ray tube 5. A structure in which the two parts 1a and 1b are joined with a fastener after the rail 4 is sandwiched vertically can be employed. Here, in an escalator provided with a step for passengers to get on between an upper floor and a lower floor as a passenger conveyor, an example in which an X-ray imaging unit is mounted on a handrail in an inclined portion is shown. The mounting location is not limited to this. At the time of X-ray imaging, the X-ray imaging unit 1 and the handrail 4 are moved relatively. That is, the X-ray imaging unit 1 is fixed and the handrail 4 is moved, or the X-ray imaging unit 1 is moved while the handrail 4 is stationary. In this way, it is possible to photograph a desired portion of the handrail. When shooting is continuously performed as a moving image and saved as a moving image file in a storage unit such as a magnetic medium of the image processing unit 2, for example, an MPEG format or an AVI format can be used.

  FIG. 5 is an example of a handrail X-ray image taken by the passenger rail handrail inspection device of this embodiment. The horizontal direction of the drawing is the longitudinal direction of the handrail. The handrail is photographed in a relatively narrow field of view of about 5 mm to 20 mm in the longitudinal direction. This is because the output of the X-ray tube is smaller than other devices such as medical devices because it does not have a cooling mechanism using oil because it needs to be compact enough to be carried by the X-ray inspection device. In order to effectively use small X-ray output, the X-ray tube is designed so that the field of view in the longitudinal direction of the handrail is about 5 mm to 20 mm and the direction perpendicular to the longitudinal direction of the handrail radiates in the range of about 50 mm Has been. However, this visual field restriction is a limitation in compact design, and the handrail X-ray inspection apparatus for passenger conveyors of the present invention can be realized using an X-ray imaging mechanism with a wider visual field. Moreover, even if the visual field in the longitudinal direction of the handrail is narrow, there is no problem because the image is taken while moving in the longitudinal direction.

  As shown in FIG. 5, the steel cord is photographed as a black line segment (dark part) because of its low X-ray transmittance. Between the steel cords arranged in a row (background), only rubber is used, and the X-ray transmittance is relatively high, so the image is bright (bright). In this example, the periphery of the handrail (upper and lower ends in the drawing) is taken dark because the rubber is thick and the X-ray transmittance is relatively low. Here, 18 steel cords are shown, and for convenience of explanation, they will be called 01 code, 02 code,.

  The luminance distribution on the left side of the drawing is a line segment in the direction of crossing the hand rail at right angles at the arrow 9 (vertical direction in the drawing, hereinafter referred to as the crossing direction or the direction perpendicular to the longitudinal direction of the hand rail). It represents the brightness distribution above. The steel cord is the valley portion of the luminance distribution. In addition, the thick part of the peripheral rubber has a low brightness and a low contrast compared to the central part.

If the X-ray imaging unit or the handrail is moved while moving at a constant speed, a single image has a field of view of about 5 mm to 20 mm, but as a movie, the entire handrail can reach several tens of meters. . Although it is desirable that the X-ray imaging unit or the handrail is moved at a constant speed during imaging, even the movement at an unsteady speed can be corrected using the encoder 3. The encoder 3 is fixed to the X-ray imaging unit, measures a relative movement amount with respect to the handrail 4, and transmits the movement amount to the image processing unit. In the image processing unit, from the moving image obtained by the X-ray imaging unit, based on the distance information of the encoder 3, if the image is collected every time a certain distance is moved, the moving image of the handrail 4 that moves at a pseudo constant speed is obtained. Can be obtained.
[Examples of good and deteriorated products]
FIG. 6 is a conceptual diagram in which an X-ray image of a non-defective handrail is imaged. For convenience of the drawing, the longitudinal direction of the handrail is shown in the vertical direction of the drawing. The steel cord part is dark (black). In this conceptual diagram, section 10 is a section in which there is a steel cord seam, and the seam is shown as bright as between steel cords with the steel cord gap slightly larger. Yes. Also, the steel cord indicates that it is meandering slightly. Such discontinuities and meandering due to the seam of the steel cord are present even in a normal handrail, and therefore cannot be determined as degradation.

On the other hand, FIG. 7A is a conceptual diagram of a deteriorated initial handrail X-ray image in which “stranding” (fraying) occurs. Since there is a discontinuity in the steel cord due to the seam described above, even if the gap is slightly large, it should not be judged as deteriorated. It is necessary to judge that the deterioration has occurred. FIG. 7B is an enlarged view of the portion where the wire is formed, and FIG. 7C is a simple schematic diagram for explaining the position of the wire. The line segment 11 is an image of a steel cord that has been turned into a bare wire.
[Shooting system]
FIG. 8 shows an example of an imaging system of a handrail inspection device for a passenger conveyor according to this embodiment.

  FIG. 8 shows the positional relationship between the X-ray exit 12 of the X-ray tube 5, the steel cord 15 with a built-in handrail (one of which is taken out), and the scintillator 6. The direction perpendicular to the drawing is the longitudinal direction of the handrail and the steel cord. A circle indicated by a solid line with reference numeral 15 shows a state in which a normal outer diameter steel cord 15 (ordinary steel cord 15) is shown from the cross-sectional direction. A state in which the steel cord 16 that is thinner than the normal state is shown from the cross-sectional direction by a circle indicated by a solid line 16. A broken line arrow 13 indicates a range irradiated with X-rays. Two line segments 17 are drawn from both ends of the X-ray exit port 12 to any point within the X-ray irradiation range on the irradiation surface of the scintillator 6. Here, the center position of the range in which the steel cord group substantially exists is represented by a broken line 4a. A circle 14 having a center on the broken line 4a and in contact with the two line segments 17 is indicated by a broken line. In the passenger rail handrail X-ray inspection apparatus of the present invention, a normal steel cord has an outer diameter larger than that of the broken-line circle 14, and a strand generated by “stranding” has an outer diameter larger than that of the broken-line circle 14. It is assumed that is small. Hereinafter, the outer diameter of the broken-line circle 14 may be referred to as the boundary outer diameter 14.

  FIG. 9 is a diagram for explaining the luminance distribution of an image formed by a normal steel cord in the photographing system of FIG. The normal steel cord 15 at the position of the broken line 4a absorbs or reflects a part of the X-rays from the X-ray exit 12 to form an image (projected image) on the scintillator 6 as a shadow picture. Two line segments 18 and two broken line segments 19 are drawn to the scintillator 6 as tangent lines of the steel cord 15 from the end of the X-ray exit 12. A region 20 (shaded region) formed on the scintillator 6 by line segments 18 and 19 drawn from different ends is a region where the steel cord 15 partially shields the emitted X-rays. That is, when the X-ray emission port 12 is viewed from above the scintillator 6 at any position in the region 20, a part of the X-ray emission port 12 is hidden behind the steel cord 15. In this way, a shadow formed by partially covering the X-ray exit 12 is called a penumbra. That is, the region 20 is a region where the steel cord 15 can be shaded. An area 21 (stipple area) outside the two line segments 19 is an area where the steel cord 15 does not cover the X-ray exit 12 at all. A region 22 between the two regions 20 is a region where the steel cord 15 completely covers the X-ray emission port 12. In this way, a dark shadow that can be completely covered by the X-ray exit 12 is called a main shadow. That is, the area 22 is an area where the steel cord 15 makes a real shadow. By forming the main shadow and penumbra, the scintillator 6 has high brightness in the region 21 and low brightness in the region 22. In the region 20, the area covered by the steel cord 15 with respect to the X-ray exit port 12 increases as the region 20 contacts the region 22 from the location in contact with the region 21, and the luminance gradually decreases.

  Further, since the X-ray image has a lot of random noise, it is necessary to apply a smoothing filter process to the original image. The luminance distribution of the scintillator 6 after the smoothing filter process is added is shown by a curve 23 in FIG. become that way. In the luminance distribution 23, the vertical axis represents luminance, and the horizontal axis represents a location on the scintillator 6. Since the X-rays that reach the scintillator 6 from the X-ray exit port 12 become weaker as they approach the periphery of the emission range, the brightness distribution 23 is precisely somewhat lower at the periphery.

  FIG. 10 is a diagram for explaining the luminance distribution of an image formed by a thin, thin steel cord in the photographing system of FIG. The steel cord 16 that is thinner than normal at the position of the broken line 4a absorbs or reflects a part of the X-rays from the X-ray exit 12 and forms an image on the scintillator 6 as a shadow picture. Two line segments 18 and two broken line segments 19 are drawn to the scintillator 6 as tangent lines of the steel cord 16 from the end of the X-ray exit 12. A region 21 (stipple region) outside the line segment 19 is a region where the steel cord 16 is not covered at all with respect to the X-ray exit 12. Further, in the case of FIG. 10, there is no main shadow because there is no region on the scintillator 6 where the steel cord 16 is completely shielded from the X-ray exit 12. The region 24 between the two line segments 18 is a range in which the area of the portion of the X-ray exit 12 hidden by the steel cord 16 is maximized when viewed from the scintillator 6. However, since the region 24 does not completely cover the X-ray exit 12 as described above, the region 24 has a higher luminance than the region 22 of the main shadow in FIG. A region 25 (shaded region) between the line segment 18 and the line segment 19 drawn from the same point at the end of the X-ray exit port 12 is a region where a part of the steel cord 16 is shielded from the X-ray exit port 12. It is. In addition, in the region 21 (stipple region) outside the two line segments 19, the scintillator 6 has high brightness because there is nothing to block the X-ray exit 12. Regions 24 and 25 are regions where the steel cord 16 forms a penumbra. In the region 25, the area where the steel cord 16 is shielded from the X-ray exit 12 increases from the position in contact with the region 21 to the position in contact with the region 24, so that the luminance gradually decreases. Further, since the X-ray image has a lot of random noise, it is necessary to apply a smoothing filter process to the original image. However, the luminance distribution of the scintillator 6 after the smoothing filter process is as shown by a curve 26. . In the luminance distribution 26, the vertical axis represents luminance, and the horizontal axis represents a location on the scintillator 6. Since the X-rays that reach the scintillator 6 from the X-ray exit 12 are weakened as they approach the periphery of the emission range, the brightness distribution 26 is precisely somewhat lower at the periphery.

  FIG. 11 is a diagram for explaining the luminance distribution of an image formed by a steel cord having a boundary outer diameter 14 in the photographing system of FIG. The steel cord having the boundary outer diameter 14 at the position of the broken line 4a forms an image as a shadow on the scintillator 6 by absorbing or reflecting a part of the X-rays from the X-ray emission port 12.

  Two line segments 18 and two broken line segments 19 are drawn to the scintillator 6 as tangent lines of the steel cord having the boundary outer diameter 14 from the end of the X-ray exit 12. A region 21 (stipple region) outside the line segment 19 is a region where the steel cord having the boundary outer diameter 14 is not covered at all with respect to the X-ray exit 12. In the case of FIG. 11, since the two line segments 18 intersect at one point on the scintillator 6, the region where the steel cord having the boundary outer diameter 14 is completely shielded from the X-ray exit 12 is on the scintillator 6. Since there is only one point, there is no substantial main shadow in the sense of having a certain range. A region outside the two line segments 19 is a region 21 (stipple region) where the X-ray exit 12 is not obstructed at all, and the scintillator 6 has high luminance. The inside of the two line segments 19 is a region 25 (shaded region) where a part of the X-ray exit 12 is shielded by a steel cord having a boundary outer diameter 14, and is a portion that causes a penumbra on the scintillator 6. . The minimum luminance value in this area is Bth_Dash. The luminance distribution 27 here is a curve that connects the background luminance Bbak, which is high luminance, to the lowest luminance value Bth_Dash. Further, since the X-ray image has a lot of random noise, it is necessary to apply a smoothing filter process to the original image. However, the luminance distribution of the scintillator 6 after the smoothing filter process is as shown by a curve 27. . In the luminance distribution 27, the vertical axis represents luminance, and the horizontal axis represents a location on the scintillator 6. Since the X-rays that reach the scintillator 6 from the X-ray exit port 12 become weaker as they approach the periphery of the emission range, the brightness distribution 27 is accurately somewhat lower at the periphery. Further, on the scintillator 6, the same darkness as that of the main shadow appears on the scintillator 6 in the drawing. Actually, however, it is discretized with pixels having a spatially finite size and the random noise described above. Bth_Dash becomes a value larger than Bmin by performing smoothing with adjacent pixels in order to remove. Then, the half-value width of the luminance distribution 27, that is, the width of the points 27a and 27b where the luminance distribution 27 crosses the intermediate luminance between Bbak and Bth_Dash is set as the steel cord thickness threshold value Wth_Dash. Using this Bth_Dash as a luminance threshold value, it is possible to distinguish a normal steel cord from a strand. As described above, Bth_Dash can be used as the brightness threshold value. However, in order to make the difference in brightness between the normal outer diameter steel cord and the strands stand out, as shown in FIG. The minimum value of the luminance value of the steel cord image having a th outer diameter (threshold outer diameter) 14a reduced by a small amount may be used as the threshold Bth.

  FIG. 12 shows a steel cord having an outer diameter (th outer diameter 14a) smaller than the boundary outer diameter 14 described in FIG. 11 and larger than the strand to be detected (wire formation) in the imaging system of FIG. It is a figure explaining the luminance distribution of the image to form. The steel cord having the outer diameter 14a at the position of the broken line 4a forms an image as a shadow on the scintillator 6 by absorbing or reflecting a part of the X-rays from the X-ray exit 12. Two line segments 18 and two broken line segments 19 are drawn as far as the scintillator 6 from the end of the X-ray exit port 12 as a tangent line of a steel cord having a th outer diameter 14a. A region outside the two line segments 19 is a region 21 (stipple region) where the X-ray exit 12 is not obstructed at all, and the scintillator 6 has high luminance. On the other hand, the region 25 and the region 24 inside the two line segments 19 are portions that cause a penumbra on the scintillator 6 when a part of the X-ray exit 12 is shielded by a steel cord having a th outer diameter 14a. is there. In a region 24 sandwiched between two points where two line segments 18 intersect with the scintillator 6, the lowest luminance is generated, which is set as a threshold value Bth. The luminance distribution 27α here is similar to the luminance distribution 27 in FIG. 11, but the minimum luminance value is higher. The half-value width of the luminance distribution 27α, that is, the width of the points 27a and 27b where the luminance distribution 27α crosses the intermediate luminance between Bbak and Bth can be used as the steel cord thickness threshold value Wth. In addition, it is possible to distinguish a normal steel cord from a strand by using Bth as a luminance threshold value.

  In the imaging system of FIGS. 8 to 12, the steel cord existing position 4 a is set to a distance close to the X-ray tube 5 for convenience of explanation. However, as shown in FIG. 13, it is possible to secure a larger field of view of the X-ray image by arranging it closer to the scintillator 6 as in the position 4b where the steel cord exists. Further, even in the case of a normal steel cord that is thicker than the boundary outer diameter 14, even in the case of a thin steel cord that has been stranded by “stranding”, the range of the regions 20, 22, 24, and 25 is on the scintillator 6. The space is compressed and becomes compact. Further, as a result, the X-ray emission range can be kept within a smaller range, and therefore, the difference in luminance that the central part of X-ray irradiation is bright and the peripheral part is dark can be reduced. On the other hand, it should be noted that the boundary outer diameter 14 for distinguishing between a normal steel cord and a thin steel cord also changes with the change of the steel cord existing position from 4a to 4b. After determining the outer diameter of the steel cord to be normal and the thin steel cord to be detected as an element wire, the boundary outer diameter 14 or the th outer diameter 14a is determined based on this, and the size of the X-ray exit 12 is further determined. By adjusting the steel cord existence position and the distance between the scintillators 6, it is possible to detect a steel cord having a desired outer diameter and a “strand”.

  As described above, according to FIGS. 8 to 13, depending on the size of the X-ray exit 12, the scintillator 6, and the positional relationship between the steel cord positions, the boundary outer diameter 14 or the th outer diameter 14 a is thicker than this. It was shown that the steel cord formed a dark shadow called the main shadow, and that the strands thinner than this formed only a relatively bright penumbra without forming the main shadow. Although FIG. 8 thru | or 13 demonstrated the case where the thin steel cord 16 as a strand was orthogonal to drawing, even when parallel on the drawing, the magnitude | size of the X-ray exit 12 and the scintillator 6 were demonstrated. And only the penumbra image due to the relationship of the steel cord existence position. Further, even if the cross section of the steel cord is not circular, if the value is not a value that approximates the boundary outer diameter 14, the image of the main shadow and the penumbra is formed so as to be easily distinguished.

The vertical and horizontal dimensions of the exit surface of the X-ray source may have an area larger than the outer diameter of the strands and the boundary outer diameter of the steel cords in order to form the main shadow of the steel cord and the penumbra of the strand. desirable.
[Image processing means of handrail inspection device]
FIG. 14 shows an embodiment of the image processing unit of the present invention, which can be implemented by the image processing unit 2, for example. The image processing unit includes a frame acquisition unit 28, a steel code detection unit (hereinafter referred to as SC detection unit) 29, a steel code model holding unit (hereinafter referred to as SC model holding unit) 30, and a steel code trace / element feature detection unit ( Hereinafter, it is composed of an SC trace / element feature detection unit) 31, a frame-by-frame quality determination unit 32, a final determination unit 33, a display unit 34, a command input unit 35, and a control unit 36. Further, the SC trace / element feature detection unit 31 includes a strand detection unit 311. Hereinafter, for convenience of explanation, the longitudinal direction of the handrail is expressed as “longitudinal direction”, and the direction orthogonal to this is expressed as “orthogonal direction”.

  The frame acquisition unit 28 cuts out one frame from the moving image taken by the camera 7, cuts out the field of view range including the steel code, and performs random noise removal and contrast correction on the image luminance.

  The SC detection unit 29 calculates a coordinate by detecting a steel code portion corresponding to a steel code existing independently from the luminance distribution of the image. An orthogonal direction of the center of gravity, that is, the coordinate in the vertical direction in FIG. 5 is represented as the position of the steel cord in the frame.

  The SC model holding unit 30 updates and holds an SC group model having a predetermined number. The SC group model updates and holds the information of the luminance of the steel code area and the luminance of the background area adjacent to the steel code area together with the coordinates of each steel code in the orthogonal direction (FIG. 29 is used for the update holding of the SC group model). Will be described later). Each steel code of the SC group model is named 01, 02,..., 18 code in FIG.

  The SC trace / element feature detection unit 31 compares the representative coordinates of the steel cord, which is the output of the SC detection unit 29, with the coordinates of the SC group model held by the SC model holding unit 30, and the total number of the predetermined number Correspondence is made so that the difference in total coordinates (total difference) is minimized. By this processing, it is determined whether each steel code detected in the current frame corresponds to the code number of each steel code, that is, the 01, 02,..., 18 code in FIG. Based on this correspondence, the presence / absence of disconnection, contact, entanglement, and stranding in the current frame of each steel code is detected and stored in the memory as an “element feature log for each frame”. The representative coordinates of each steel code are used by the SC model holding unit 30 for the update information of the SC group model. Then, each steel code coordinate of the updated SC group model is stored in the memory as a “trace log for each frame”.

  The frame pass / fail judgment unit 32 refers to the “element feature log for each frame”, and the element features occurring in the current frame, that is, the presence or absence of steel cord omission, contact, entanglement, stranding, and these Check how many frames the feature has continued, and record “bad” in the pass / fail judgment log for each frame if “wired” continues for more than a predetermined frame, but not continue Records “good”. In the case of “defect”, the process goes back to the frame in which the element feature occurs, and records “defect” in the log in the meantime. In this manner, the “good / bad determination log for each frame” is updated and stored in the memory. Here, it is also possible to store a multi-step pass / fail judgment as a log by using a logical product result of the presence / absence of a plurality of element features.

  The final determination unit 33 refers to the entire “good / good determination log for each frame”, determines the quality of the entire handrail sample, and outputs the final determination result.

  The command input unit 35 is realized by a well-known command input device such as a keyboard and a mouse, and can control the start and stop of the processing of the image processing unit 2 via the control unit 36.

  The display unit 34 can use a known graphical user interface such as a PC display. It can be configured such that the quality determination result output from the final determination unit 33 is displayed on the display unit 34 together with the image of the camera 7.

The image processing unit 2 of the present invention can be obtained only by inputting the same result as the result output from the final determination unit 33 to the command input unit 35 in order to call attention of maintenance personnel using the handrail inspection device of the present invention. It is also possible to configure so that the above process can be terminated. The configuration of the image processing unit 2 has been described above.
[Frame acquisition means]
The frame acquisition unit 28 sequentially extracts and processes frames from the acquired moving image. This processing may be configured to perform on-line processing, ie, sequential processing each time a frame is received from the camera 7, or off-line processing, ie, temporarily store the entire measured portion of the handrail 4 as a moving image file. You may comprise so that it may read and process, after storing in an apparatus. The off-line processing can be performed without imposing an excessive load on the image processing unit 2.

  When the X-ray imaging unit 1 or the handrail 4 moves at an unsteady speed and performs online processing based on the distance information of the encoder 3, the frame acquisition unit 27 determines whether the handrail 4 is in accordance with the output of the encoder 3. For example, every time a predetermined distance between 5 mm and 20 mm moves, a frame is acquired and processed. In the case of offline processing, a frame is acquired every time the handrail 4 moves, for example, a predetermined distance between 5 mm and 20 mm in accordance with the output of the encoder 3, and a thinned video file is configured to be temporarily stored. Can be configured to be stored in a magnetic storage device. The moving image file created in such a configuration is a handrail moving image that moves at a constant speed. The frame acquisition unit 28 can read this moving image, acquire it one frame at a time, and process it. When performing offline processing, the function of saving the acquired image as a moving image file in the frame acquisition unit 28 is not essential, and moving image generation may be performed separately by moving image generation software.

  Here, the relationship between the moving speed of the handrail 4, the visual field width in the handrail longitudinal direction of the handrail X-ray image shown in FIG. When the frame rate of the camera 7 is N frames per second, the time interval of frames output from the camera 7 is 1 / N second. On the other hand, if the moving speed of the handrail 4 is L millimeters per second, the camera 7 is moving L / N millimeters while acquiring one frame. Therefore, if it is assumed that the shutter of the camera 7 is open, if the visual field length in the longitudinal direction of the handrail 4 is larger than L / N millimeters, all the image information of the handrail 4 is stored in the moving image file. include.

  For example, when the frame rate of the camera 7 is 30 frames per second and the moving speed of the handrail 4 is 500 millimeters per second, the movement amount during acquisition of one frame is 500 / 30≈16.7 millimeters. Therefore, if the visual field length of the handrail X-ray image is 17 millimeters or more, all the image information of the handrail 4 is included in the moving image file.

Further, when an interlaced camera is used as the camera 7, each frame obtained at 30 frames per second includes two images having an acquisition time of about 16.7 milliseconds as an even field and an odd field. If these are separated and processed as two images, an image can be obtained substantially at a frame rate of 60 frames per second. In this case, the movement amount during the acquisition of one frame is 500 / 60≈8.3333 millimeters. Therefore, if the size of the handrail X-ray image shown in FIG. 5 in the longitudinal direction of the handrail is 8.4 millimeters or more, all the image information of the handrail 4 is included in the moving image file.
[SC detection means]
Next, the SC detection unit 29 will be described. FIG. 15 is a diagram showing a first example of processing performed by the SC detection means, and shows the flow of processing. As shown in FIG. 5, for example, the X-ray image by the frame acquisition unit 28 is projected in the longitudinal direction of the handrail to create a projection luminance distribution (S1). By projecting, it is possible to cancel random luminance noise associated with X-ray imaging, and to measure the tendency of luminance according to the location of the handrail based on the difference in rubber thickness. A contrast correction curve is created from the projection profile of S1 (S2). Next, the HR orthogonal direction line is set to the left end in the longitudinal direction in order to analyze the luminance distribution in the orthogonal direction of the X-ray image while skipping line by line or predetermined lines from the left end to the right end (S3).

  Thereafter, a loop process (S4) is performed in the longitudinal direction of the handrail (indicated as HR in FIG. 15), and the X-ray image is analyzed within the visual field range by the frame acquisition unit 28. That is, in the set line, a brightness distribution obtained by smoothing and contrast correcting the acquired brightness distribution is created (S5), and an SC candidate cell of the set line is detected (S6). An SC candidate cell is a pixel that may be part of a steel code. Details of S6 will be described later. In the first loop of the S4 loop, S7 passes without processing, but after the second loop, the SC candidate cells detected in the previous loop and the SC candidate cells detected in the current loop are connected. (S7). How to connect will be described later. A line on the right side of the line analyzed in the current loop is set (S8). When the analysis is completed up to the right end of the X-ray image, the process exits the S4 loop (S9).

  When the connection length of the SC candidate cell detected in the S4 loop reaches a predetermined length, it is determined as a steel code, and the coordinates in the orthogonal direction among the barycentric coordinates of the connection object of the SC candidate cell are determined in the frame. (S10). In S10, the determined number of steel codes and the orthogonal coordinates of the determined steel code in the frame are output, and the process of the SC detection unit 29 is completed.

  Next, the process of S5 will be specifically described. FIG. 16 is a diagram illustrating an example of contrast correction of the SC detection unit 29. FIG. 16A shows the luminance distribution on a certain line in the orthogonal direction of the handrail X-ray image acquired by the frame acquisition unit 28. The vertical axis of the drawing is the coordinate in the direction perpendicular to the handrail, and the vertical axis of the drawing is the luminance. The luminance distribution becomes valleys at the steel cords and peaks between the steel cords. Moreover, there are small peaks and valleys in the luminance distribution due to the random noise of luminance accompanying X-ray imaging. Further, the luminance is low near the upper end and near the lower end of the drawing, and the contrast between peaks and valleys is poor. Smoothing and contrast correction (S5) are performed on these to reveal the peaks and valleys resulting from the steel cord.

  FIG. 16B shows the result of smoothing the luminance distribution of FIG. Smoothing can be realized by, for example, a known smoothing filter described in Non-Patent Document 1. As described above, random noise can be suitably removed by using a smoothing filter having a sufficiently large size so as not to eliminate the peaks and valleys caused by the steel cord.

  If imaging system parameters such as the lens of the camera 7 are determined, the number of pixels on the image of the steel code to be detected is almost determined, so that an appropriate size of the smoothing filter can be determined in advance. In addition, the vicinity of the upper end and the lower end of the drawing corresponding to the periphery of the handrail is corrected because the contrast between the brightness and the valley is low.

  FIG. 16C shows a longitudinal projection luminance distribution by S1. A broken line 37 is an envelope of the maximum value of the longitudinal projection luminance distribution. After calculating the primary difference of the luminance distribution and detecting the vicinity of the zero crossing of the primary difference as a plurality of maximum values, a curve in contact with the plurality of maximum values can be obtained by, for example, Lagrange polynomial approximation. When the broken line 37 thus obtained is set to f (x) and, for example, CONST / f (x) is multiplied by the smoothed luminance distribution (b), a luminance distribution with corrected contrast is obtained (FIG. 16 (d)). Here, x represents coordinates in the orthogonal direction, CONST is a constant, and when the luminance value or luminance distribution value of the image is from 0 to 255, approximately 200 is an appropriate value. Since the broken line 37 which is the value of f (x) is used for division, it is also effective to perform clipping so as not to become a numerical value of 10 or less so as not to become too small so as not to become unstable. Clipping in this case is a process of replacing with a numerical value 11 or the like larger than the lower limit when a numerical value of 10 or less appears. FIG. 16D is a luminance distribution obtained by contrast correction with respect to FIG. 16B, and there are 18 sharp valleys, which correspond to 18 steel cords. If a sharp valley can be formed in this way, a SC candidate cell can be set at a point where this distribution is first-order-differed and zero-crossed.

  Next, the steel code (SC) candidate cell detection S6 and the SC candidate cell concatenation S7 are specifically shown. FIG. 17 is a diagram illustrating an example of SC candidate cell detection (S6) by the SC detection unit 29. FIG. 17A shows the luminance distribution after contrast correction as in FIG. 16D, and FIG. 17B shows the distribution obtained by linearly subtracting the luminance distribution in the orthogonal direction. The primary difference is a value obtained by subtracting the luminance value below the attention point and the luminance value above the attention point. The distance (number of pixels) from the attention point to the upper and lower sides is appropriately adjusted in advance. It is necessary to keep it. When the outer diameter of the steel cord to be detected is generally determined as in the present invention, the adjusted distance can be held as a fixed value. The positive threshold value 38 and the negative threshold value 39 can also hold the adjusted appropriate values fixedly.

  For example, the portion corresponding to the valley 40 of the luminance distribution in FIG. 17A is a section from when the primary difference value in FIG. 17B increases from the negative threshold value 39 to the positive threshold value 38. What is necessary is just to obtain | require the coordinate which takes the minimum value of the luminance distribution of Fig.17 (a) in the coordinate area of the orthogonal direction applicable to 41 and 42. FIG. The sharpness of the valley 40 is calculated by calculating the distance in the orthogonal direction between the maximum value 43 and the minimum value 44, and can be selected as a sharp valley when the distance is equal to or less than a predetermined fixed value. The location selected as a sharp valley in this way is the detection position of the SC candidate cell in the luminance distribution of FIG.

  FIG. 18 is a diagram illustrating the SC candidate cell concatenation (S7) process of the SC detection unit 29. FIGS. 18A, 18 </ b> B, 18 </ b> C, and 18 </ b> D are a part of an image to be processed by the SC detection unit 29. The hatched areas 45 and 46 are steel cords that exist independently, the hatched area 47 is in a state where two steel cords are in contact, and the hatched area 48 is a short steel cord. A broken line group 49 in the vertical direction of the drawing explicitly represents the luminance distribution analysis lines set in S3 and S8. That is, the luminance distribution is analyzed along these lines, and SC candidate cells are determined in the valley portions.

  In FIG. 18B, a white rectangle 50 on the hatched area 45 is one of SC candidate cells. Other rectangles on the hatched area 45 are also SC candidate cells, but are not numbered because they become complicated. White rectangles on the hatched areas 46 and 48 are also SC candidate cells. In the hatched area 47, no sharp valleys are generated in the luminance distribution due to the contact, and therefore an SC candidate cell cannot be set. The fact that no SC candidate cell can be set means that there is no steel code that exists independently without touching.

  FIG. 18C explicitly shows the operation of the S4 processing loop. That is, the luminance distribution analysis lines 49a,..., 49d are processed from the left end to the right end, and the SC candidate cells that have been set are solid white rectangles. The dotted line is shown. In S7, the SC candidate cell set in the previous line and the SC candidate cell set in the current line are connected. In the connection, SC candidate cells having the shortest distance are connected. The distance is a difference in orthogonal direction coordinates between SC candidate cells. If the upper limit of the distance that can be connected is determined, incorrect connection can be prevented.

  The solid lines 51, 52, and 53 in FIG. 18D explicitly show the result of connecting SC candidate cells in S7. This indicates that there are independent long steel cords 51 and 52 and a short steel cord 53 in the frame.

  In S10, a length threshold value is determined in advance, a connection result of a predetermined length or more is determined as a steel cord, and a connection result less than the threshold value is removed. The handrail X-ray image acquired by the frame acquisition unit 28 has an image size in the longitudinal direction fixed when the X-ray imaging unit 1 is designed, so that the threshold value of the length can be determined.

  In the above description, the smoothing and the contrast correction are performed every time one line is analyzed in the HR processing loop S4. However, a two-dimensional image is processed as in the processing flow of the second processing example of the SC detection unit shown in FIG. After performing smoothing and contrast correction processing as processing (S11), the luminance distribution analysis for each line may be performed. In FIG. 19, the processing of S1 and S2 is the same as the processing described in FIG. In S <b> 11, luminance random noise associated with X-ray imaging is canceled by applying a two-dimensional smoothing filter in the handrail X-ray image acquired by the frame acquisition unit 28. Smoothing can be realized by a known two-dimensional smoothing filter described in Non-Patent Document 1. In contrast correction, the pixel in the X-ray image is multiplied by CONST / f (x) using the contrast correction curve f (x) obtained in S2. At this time, if the pixel value exceeds an upper limit such as 255, clipping processing such as replacement with the upper limit value is performed. In the method described above, S11 creates an X-ray image that has been smoothed and subjected to contrast correction.

S3 and S4 are similar to the processing described with reference to FIG. 15, and the luminance distribution analysis is performed while skipping one line or a predetermined line from the left end to the right end of the X-ray image. The luminance distribution obtained here is equivalent to the luminance distribution obtained in S5 of FIG. 15, that is, FIG. The subsequent processing from S6 to S10 is the same as the processing flow described in FIG.
[SC model holding means]
FIG. 20 is a diagram showing a method for collating the SC model with the detected steel code. Next, referring to FIG. 20, a method of collating the SC model with the detected SC, which is performed by the cooperative processing of the SC model holding unit 30 and the SC trace / element feature detection unit 31, will be described.

  To simplify the description, it is assumed that the number of steel cords built in the handrail is originally five. Then, the SC model is also composed of five steel codes, and can be named 01 code, 02 code, 03 code, 04 code, 05 code in order from the end.

  Here, it is assumed that five steel cords are detected by the SC detection unit 29. In this case, the steel cord detected in order from the end unconditionally is associated with the steel cord of the SC model (FIG. 20A). That is, the detected steel codes 54, 55, 56, 57, and 58 are associated with the 01 code, 02 code, 03 code, 04 code, and 05 code of the steel code model, respectively. By this process, the names 01 to 05 are determined for each detected steel code. Each code of the SC model has position information (coordinates), and the coordinates of the SC model are updated with the coordinates of the center of gravity representing the position of the steel code detected and associated. The white circle at the center of the detected steel cord in FIG. 20A explicitly indicates the position of the center of gravity.

  FIG. 20B shows a case where three steel cords are detected. This is a case where the other two cords are missing or cannot be detected as two independent steel cords due to contact with each other. If a restriction condition is provided that the order of the steel codes of the SC model is not changed up and down, it is 01 code, 02 code, and 03 code that the detected steel code 59 is associated with the SC model. If the code is associated with the 04 code, one of the detected steel codes 60 and 61 is left over. Under the same constraint conditions, the detected steel code 60 is associated with the SC model in the 02 code, 03 code, and 04 code, and for the detected steel code 61, the 03 code, 04 code, and 05 code. Is associated with.

  The distance table in FIG. 20C summarizes these conditions in a table. The column at the left end represents the SC model code name. The first line, which is the upper end, is a steel code detected in the current frame, and is written from left to right in order from the smallest detected coordinate. In this example, the steel cords 59, 60, 61 are sequentially detected from the left. Since it is unclear at this point which steel code has been lost, it is not possible to determine the name of the detected steel code. In the distance table, the absolute value of the difference between the coordinates held by the model and the coordinates of the detected steel cord is entered at the intersection of the row and column. A combination in which an asterisk is entered is a combination that is not possible from the above-described constraint conditions. Other locations, that is, d11, d21, d31, d22, d32, d42, d33, d43, and d53 are absolute values of coordinate differences. The vertical column at the right end of the distance table is a column for entering the minimum coordinate difference which is the minimum value of the coordinate difference. The 01 code and 05 code are entered because they only have the values of d11 and d53, respectively. For other parts, enter the minimum value of the corresponding line. For example, in the 02 code line, the smaller of d21 and d22 is entered. If both are equal, enter the number.

  Here, since there are three steel cords detected with respect to five steel cords of the SC model, any two of them are not detected. Therefore, two are selected in order from the largest column in the right end of the distance table. The steel code corresponding to the line selected here is regarded as undetected. For example, when the 02 code line and the 03 code line are selected, the 02 code and 03 code are not detected. As a result, the detected steel code 59 is associated with the 01 code of the SC model and is determined to be the 01 code. The detected steel code 60 is associated with the SC code 04 code and is determined to be 04 code. The detected steel code 61 is associated with the 05 code of the SC model and is determined to be the 05 code.

  With the above processing, the name of the detected steel code is determined. On the other hand, the coordinates of the 01 code of the SC model are updated with the coordinates of the steel code 59, and the coordinates of the 04 code of the SC model are updated with the coordinates of the steel code 60. The coordinates of the 05 code of the SC model are updated with the coordinates of the steel code 61. The SC model steel codes 02 and 03 are updated between the 01 and 04 coordinates with the coordinates internally divided by 1: 2 and 2: 1, respectively.

  The process described above is performed according to the process flow of FIG. FIG. 21 is a diagram showing a processing flow of the SC model holding unit 30 and the SC trace / element feature detection unit 31. If the number of detected steel cords is equal to the design number of steel cords built in the handrail, it is determined that all the steel cords have been detected, and S21 is executed to end the processing here. In S21, the codes of the steel model code name of the SC model are assigned in ascending order from the smallest coordinate of the detected steel code. The steel code information of the SC model is the coordinates of the corresponding steel code and the brightness of the original image, that is, the handrail X-ray image acquired by the frame acquisition unit 28 at the coordinate position. This is the luminance value of the corresponding steel cord part. In addition, the luminance of places on both sides of the region and separated by a predetermined distance is held as the background luminance. The predetermined distance is a midpoint between adjacent steel cord coordinates. In S21, these coordinate values and luminance values are updated.

  On the other hand, if the number of detected steel cords is insufficient, the process proceeds to process P1 (S20). In process P1, the distance table described with reference to FIG. 20C is created for the steel cord detected as the current SC model (S22). As described above, an undetected steel code is identified based on the minimum coordinate difference, and a code name is assigned to the detected steel code (S23). The coordinates of the SC model are updated based on the detected coordinates of the steel cord (S24). At this time, the luminance value is not updated.

Heretofore, the steel code tracing and the SC model updating by the cooperative processing of the SC model holding unit 30 and the SC trace / element feature detection unit 31 have been described. The SC trace is performed by the SC trace unit of the SC trace / element feature detection unit 31. In the explanation so far, there has been no case where more steel cords than the design number of steel cords are detected. This is because the SC detection unit 29 sufficiently performs smoothing processing and the like so as not to detect a fake steel code, so there are no cases where many detections are made.
[Element feature detection method]
Next, the element feature detection method related to the steel code deterioration performed by the SC trace / element feature detection unit 31 will be described. Here, detection is performed by the element feature detection unit of the SC trace / element feature detection unit 31. The element feature is an appearance feature related to deterioration recognized in the frame, and includes a lack (disconnection) of steel cords, contact between steel cords, and the presence of a bare steel cord. In addition, the contact is not limited to the adjacent steel cords, and it is an entanglement feature when non-adjacent steel cords or three or more steel cords are in contact.

  FIG. 22 is a diagram showing a method for detecting an element feature “missing” in the SC trace / element feature detecting unit 31. In FIG. 22, a rectangle 62 having 16 black horizontal stripes is a conceptual diagram of a handrail X-ray image. The 16 black horizontal stripes represent 16 steel cords that exist independently, and these are detected by the SC detection unit 29. A horizontally long rectangle surrounded by a broken line 63 indicates the detected steel cord. Further, 18 horizontal line segments 64 indicate the positions of the steel model coordinates of the SC model held by the SC model holding unit 30, and are 01 code, 02 code,..., 18 code in order from the top. The 16 steel codes 63 detected by the cooperative processing of the SC model holding unit 30 and the SC trace / element feature detection unit 31 described above have been named. In this example, 09 code and 10 code are not detected. The coordinates where the undetected steel code should be can be known from the SC model, and are the area of the broken line 65.

  Therefore, when the minimum luminance of the area of the broken line 65 in the X-ray image is calculated and is larger than a predetermined threshold value, it is determined that the steel cord is missing. As the predetermined threshold value, the average value of the luminance of the corresponding steel code of the SC model and the background luminance adjacent thereto can be used. In this example, the average value of the 09 code and 10 code of the SC model is the luminance of the SC model, and the luminance at the midpoint coordinates of the coordinates of the 09 steel code and the 10 steel code is the background luminance of the SC model. And the average value of the background luminance is set as the predetermined value, which is used as a threshold for determining whether or not the steel cord is missing. As described above, the luminance from the 01 code to the 18 code of the SC model and the background luminance between them are values updated in the frame in which all the steel codes are detected. It can be determined that the steel cord is absent if the steel cord is absent and the brightness is high at a place where the steel cord is to be detected independently. By the process described above, the absence of the steel cord is detected in the frame.

  FIG. 23 is a diagram showing a method of detecting the element feature “contact” in the SC trace / element feature detection unit 31. In FIG. 23, a rectangle 66 having 17 black horizontal stripes is a conceptual diagram of a handrail X-ray image. In the region of the broken line 69 is a steel cord that has become thicker due to contact. The other 16 black horizontal stripes represent 16 steel cords that exist independently, and these are detected by the SC detector 29. A horizontally long rectangle surrounded by a broken line 67 indicates the detected steel cord. A steel cord having a thick appearance as indicated by the broken line 69 cannot be detected because a sharp valley does not appear in the luminance distribution as described in the description of the SC detection unit 29. Further, 18 horizontal line segments 68 indicate the positions of the steel model coordinates of the SC model held by the SC model holding unit 30, and are 01 code, 02 code,..., 18 code in order from the top. By the cooperative processing of the SC model holding unit 30 and the SC trace / element feature detection unit 31 described above, 16 detected steel cords 67 have been named. In this example, 09 code and 10 code are not detected. The coordinates where the undetected steel code should be can be known from the SC model, and are the area of the broken line 69.

  Therefore, the minimum luminance in the region of the broken line 69 in the X-ray image is calculated, and if it is smaller than a predetermined threshold value, it is determined that a plurality of steel cords are in contact. As the predetermined threshold value, as described above, the average value of the luminance of the corresponding steel code of the SC model and the background luminance adjacent thereto can be used. In this example, since the steel cord 09 and the steel cord 10 are not detected, it is determined that these steel cords are in contact with each other. It can be determined that a plurality of steel cords are in contact if the steel cord is absent and the brightness is low at a place where the steel cord is to be detected independently. By the processing described above, the contact of the steel cord is detected in the frame.

  FIG. 24 is a diagram showing a method for detecting an element feature “entanglement” in the SC trace / element feature detector 31. In FIG. 24, a rectangle 70 having 16 black horizontal stripes is a conceptual diagram of a handrail X-ray image. In the region of the broken line 73 is a steel cord that has a thick appearance due to entanglement. The other 15 black horizontal stripes represent 15 steel cords that exist independently, and these are detected by the SC detector 29. A horizontally long rectangle surrounded by a broken line 71 indicates the detected steel cord. A steel cord having a thick appearance as indicated by a broken line 73 cannot be detected because a sharp valley does not appear in its luminance distribution as described above. The 18 horizontal line segments 72 indicate the positions of the steel model coordinates of the SC model held by the SC model holding unit 30, and are 01 code, 02 code,..., 18 code in order from the top. By the cooperative processing of the SC model holding unit 30 and the SC trace / element feature detection unit 31 described above, 15 detected steel cords 71 have been named. In this example, the 08 code, 09 code, and 10 code are not detected. The coordinates where the undetected steel code should be can be known from the SC model, and are the area of the broken line 73.

  Therefore, when the minimum luminance of the area of the broken line 73 in the X-ray image is calculated and is smaller than a predetermined threshold value, it is determined that a plurality of steel cords are in contact. As the predetermined threshold value, as described above, the average value of the luminance of the corresponding steel code of the SC model and the background luminance adjacent thereto can be used. In this example, since the steel cord 08, the steel cord 09, and the steel cord 10 are not detected, it is determined that these steel cords are in contact with each other. In this case, since three or more steel cords are in contact with each other, and steel cords other than the adjacent steel cords are also in contact, it is determined that entanglement has occurred.

By the way, in order to refer to the luminance of the broken line 65 region, the broken line 69 region, and the broken line 73 region in order to determine the presence / absence of missing, touched and tangled steel cords, the minimum luminance of the region in the X-ray image is used. Instead, the minimum luminance of the region in the smoothed image obtained by the SC detection unit 29 may be used. If comprised in this way, it will not be influenced by the dispersion | variation in the brightness | luminance at the time of X-ray imaging.
[Wire detection method in element feature detection method]
Next, the strand detection unit 311 and the strand detection method for detecting the strands that appear by “stranding” included in the SC trace / element feature detection unit 31 will be described.

  FIG. 25 is a conceptual diagram of a handrail X-ray image, and shows an example in which a part of a steel cord has been turned into a strand.

  In FIG. 25, black rectangles 74 to 77 are typical examples of images of steel cords having a normal outer diameter. The hatched linear regions 78 and 79 are examples of strands (fraying) to be detected here. The strand is generally about 1/10 to 1/5 of the outer diameter of a normal steel cord, but as described with reference to the imaging system with reference to FIGS. 8 to 12, the size of the X-ray exit 12 is small. Since it is finite and not zero, only a penumbra is formed for the strands thinner than the boundary outer diameter 14 or the th outer diameter 14a. Therefore, the strands 78 and 79 are not only narrower than the normal steel cords 74 to 77 but also have a high luminance. Further, the longitudinal direction of the wire itself may be not only in the longitudinal direction of the handrail, but in any direction. Moreover, since the peripheral part of the steel cord which has a normal outer diameter is a penumbra, it takes the brightness | luminance near a strand. However, since the position of the steel code detected by the SC detection unit 29 is a valley portion of the luminance distribution for detection, it is a portion corresponding to the main shadow. Therefore, the luminance of the steel cord having a normal outer diameter (normal steel cord) at the position detected by the SC detection unit 29 is the luminance of the main shadow, that is, Bmin defined by the luminance distribution 23 of FIG.

  Here, the brightness of the image formed by the normal (ordinary) steel cord and the strand is discussed. As described above, a normal steel cord forms a dark main shadow, but a strand thinner than the boundary outer diameter 14 becomes an image of only a penumbra. The absolute brightness of the main shadow and the penumbra varies depending on the sensitivity of the scintillator 6 and changes on the image captured by the camera 7 according to the sensitivity and gain adjustment of the camera. Therefore, after determining the size of the X-ray tube 5, scintillator 6, steel cord presence position 4 a, and X-ray exit 12 of the imaging system described with reference to FIGS. 8 to 12, the gain of the camera 7 is determined. To do. Thereafter, a normal steel cord is photographed by the camera 7, and the luminance value of the main shadow created by the steel cord in the photographed digital image is defined as Bmin. When the luminance of one pixel is quantized to 0 to 255, it is desirable to adjust Bmin to be 80 or less, and an area that is neither a main shadow nor a penumbra, that is, an area 21 that is a background. It is desirable to adjust the luminance Bbak of the portion to 160 or more.

  Next, an image formed by the object having the boundary outer diameter 14 or the th outer diameter (threshold outer diameter) 14a is taken by the camera 7, and a penumbra made by the boundary outer diameter 14 or the th outer diameter 14a in the photographed digital image. Let Bth be the lowest luminance in the.

  Here, the outer diameter of the boundary outer diameter steel cord having an outer diameter that becomes a boundary between the normal (normal) steel cord and the outer diameter of the strand will be discussed. When discussing the outer diameter of a steel code or wire on an image, it is convenient to discuss the number of pixels when viewed as a digital image. 11 or 12, the X-ray tube 5, the scintillator 6, the steel cord existence position 4a, the size of the X-ray exit 12 and the steel cord having the boundary outer diameter 14 or the th outer diameter 14a are used. The luminance distribution 27 or 27α is plotted. Since it is preferable to use the th outer diameter 14a rather than the boundary outer diameter 14, hereinafter, the case where the th boundary outer diameter 14a is used will be described with reference to FIG. Regarding the luminance distribution 27α, the curve of the transition section from the lowest luminance value Bth to the background (the region that is neither the main shadow nor the penumbra, that is, the region 21) luminance value Bbak may be connected by a straight line at both ends of the interval. Then, the boundary width Wth is determined from the difference between the locations 27a and 27b when the luminance distribution 27α crosses the intermediate value Bmdl (for example, Bmdl = (Bth + Bbak) / 2) between Bth and Bbak of the luminance distribution 27α. The predetermined width determined based on the boundary width Wth is originally in the range of 1/10 mm to several mm, but may be divided by the pixel size per pixel when photographed by the camera 7 to obtain the number of pixels. it can. In the future, the boundary width Wth is the number of pixels determined based on the th outer diameter 14a.

  Based on the above, FIG. 1 shows a processing flow for detecting a strand represented by the strand 78 and the strand 79. In FIG. 1, the entire screen of the X-ray image obtained by the frame acquisition unit 28 is smoothed (S30). This is to remove random luminance change noise which is a feature of the X-ray image. The smoothing process can be achieved by, for example, a 3 × 3 pixel smoothing filter. This is achieved by replacing the average value of the neighboring 8 pixels adjacent in the vertical and horizontal directions and diagonally with the luminance of the target pixel. This smoothing can also be realized by a known smoothing filter as shown in Non-Patent Document 1. In order to execute the processing loop of S32 on the smoothed image, the image processing unit 2 sets a pointer (scanning start point) for accessing the pixel value at the upper left of the image (S31). The luminance value (pixel value) of the pixel of interest is compared with Bth, which is the lower limit of the penumbra luminance value formed by the steel code of th outer diameter 14a. If it exceeds Bth, line segment detection processing S35 described later is performed (S34). . This is because the luminance of the penumbra formed by the strand to be detected exceeds Bth as described above. When the boundary outer diameter 14 is used, Bth_Dash may be used as the Bth value. It is determined whether or not the processing range of the image has been ended by the line segment detection processing S35, and if it has ended (true), the processing loop S32 is exited (S36). If not completed, the pixel pointer is updated (S37), and finally the line segment detection process S35 is performed for the processing range of the image. In the line segment detection processing S35 described later, a processing kernel having a predetermined size such as 7 × 7 pixels is used, but it should be noted that the processing kernel does not protrude from the processing range in the processing of S35. For example, when the size of the image is 640 pixels wide and 480 pixels long and the processing range is the entire surface, the processing by the 7 × 7 pixel kernel performs the pixel pointer from the 4th pixel to the 637th pixel in the horizontal direction. Is updated (S36, S37). In the vertical direction, the pixel pointer is updated from the fourth pixel (column) to the 477th pixel (column) (S36, S37). Next, a line segment analysis process S33 described later is executed on the processing result of the processing loop S32 to detect a strand.

  The line segment detection process S35 is a process for detecting a line segment in any direction that is equal to or larger than the penumbra luminance value lower limit Bth and equal to or smaller than the intermediate luminance value Bmdl serving as a reference for the boundary width Wth. The all directions may be eight neighboring directions, that is, a horizontal direction with respect to the screen, an upper left oblique 45 ° direction, an upper right oblique 45 ° direction, a vertical direction, or a finer pitch. In the case of the vicinity of 8, for example, as shown in FIG. 2, the line segment detection filters a1, a2, a3, and a4 are sequentially applied to pick up pixels that meet a predetermined condition, and the result is displayed in the line segment total image 80 (described later). Tally. It is assumed that all pixels of the line segment total image 80 are cleared to zero prior to the line segment detection process s35. Here, the filter is 7 × 7 pixels, but 3 × 3 pixels, 5 × 5 pixels, 9 × 9 pixels, and the like can be selected in accordance with the boundary width Wth. For example, the line segment detection filter a1 detects a penumbra extending in the horizontal direction as indicated by b1, and the reference pixel m1 of the line segment detection filter a1 is aligned with the target pixel. That is, the luminance of the reference pixel m1 is the luminance value of the target pixel. Here, reference is made to the luminance values of the pixels corresponding to ref11 and ref12, here the upper and lower third pixels of the target pixel, and if the luminance of these reference pixels is higher than the reference pixel m1, the position of the target pixel is determined. It is assumed that there is a penumbra b1 extending in the horizontal direction. Then, the predetermined pixel value of the line segment total image 80 is set to 255, for example, so that it can be identified.

  Similarly, the reference pixel m2 of the line segment detection filter a2 is matched with the pixel of interest and compared with the pixel values corresponding to ref21 and ref22. If the luminance of the pixel value of interest is lower than these, the upper left is located at the position of the pixel of interest. It is assumed that there is a penumbra b2 extending in an oblique 45 ° direction. Then, the predetermined pixel value of the line segment total image 80 is set to 255, for example, so that it can be identified.

  Similarly, the reference pixel m3 of the line segment detection filter a3 is aligned with the target pixel and compared with the pixel values corresponding to ref31 and ref32. If the brightness of the target pixel value is lower than these, the position of the target pixel is It is assumed that there is a penumbra b3 extending in an oblique 45 ° direction. Then, the predetermined pixel value of the line segment total image 80 is set to 255, for example, so that it can be identified.

  Similarly, the reference pixel m4 of the line segment detection filter a4 is matched with the pixel of interest and compared with the pixel values corresponding to ref41 and ref42. If the luminance of the pixel value of interest is lower than these, it is perpendicular to the position of the pixel of interest. It is assumed that there is a penumbra b4 extending in the direction. Then, the predetermined pixel value of the line segment total image 80 is set to 255, for example, so that it can be identified. The reason that the luminance value of each standard pixel is lower than that of the reference pixel such as ref11 is that the penumbra is darker than the background. Although this condition can occur even on the main shadow of a normal steel code, since the process of selecting only pixels having luminance exceeding the luminance Bth (S34) is performed, erroneous detection of the main shadow is suppressed. In addition, there may be pixels having luminance exceeding the luminance Bth on the normal steel code penumbra, but these do not meet the condition of the line segment filters a1 to a4 that both sides of the target pixel have higher luminance. False detection is suppressed. Through the above processing, the line penumbra candidates are aggregated in the line segment aggregate image 80 as a binary image.

  FIG. 3 is a diagram illustrating an example of a line segment total image, and FIG. 3A is an example of an X-ray image. FIG. 3B is a conceptual diagram showing an example in which the strands 81 to 85 appear. FIG. 3C shows a line segment total image 80 totaled by the processing loop S32.

  Here, since the line segment total image 80 is a binary image, in the line segment analysis process S33, by removing the isolated points or after the labeling process, the line segments less than the predetermined pixel are excluded. It is possible to remove false alarms that could not be excluded in the previous stage and trivial strands that can be ignored.

  Further, in the line segment analysis processing S33, when there is a pixel or label relating to the detection line segment obtained by the above processing, a strand due to “stranding” occurs in the image (the frame in the moving image). It is determined that Incidentally, the line segment detection filter shown in FIG. 2 is an example of the line segment detection filter, and the well-known edge detection filter described in Non-Patent Document 2 can also be used.

Through the processing described above, the steel cord is detected to be disconnected, touched, entangled, or stranded in the frame. Each time frame processing is performed, the presence or absence of element features such as missing, contact, entanglement, or stranding in the frame is retained and updated as a history for each feature. This element feature history is called an “element feature log for each frame”, which is an output of the SC trace / element feature detector 31. Further, the update history of the coordinates of each steel code of the SC model described above can be updated and held for each frame. The update history of the coordinates of each steel code of this SC model is called “trace log for each frame”, which is also an output of the SC trace / element feature detection unit 31.
[Failure judgment means for each frame]
FIG. 26 is a diagram showing an example of pass / fail judgment conditions in the pass / fail judgment unit 32 for each frame. The frame pass / fail judgment unit 32 refers to the “element feature log for each frame”, judges the pass / fail of the steel cord built in the handrail to be inspected in each frame, and records the history of each pass / fail judgment frame. Keep updated. This frame-by-frame history is referred to as a “frame-by-frame quality determination log”, which is an output from the frame-by-frame quality determination unit 32.

  Whether the steel cord is good or bad is determined by using as an index the length that the element feature of “stranding” (fraying) continues in the longitudinal direction of the steel cord. For example, when a predetermined natural number of Sth is set as a longitudinal threshold value of “stranding” and continues beyond this threshold value, it is determined as deterioration (or failure), and the determination is made as “ It is stored in the “good / bad determination log for each frame”. Further, in the case of deterioration (or failure), the process goes back to the frame where the element feature “stranding” has occurred, and records all deterioration (or failure) in the log in the meantime. In this manner, the “good / bad determination log for each frame” is updated and stored in the memory. In addition, when it does not correspond to deterioration (or failure) (when it does not continue), it is recorded as a non-defective product in the “quality determination log for each frame” of the frame.

  This determination can be performed, for example, as shown in FIG. In FIG. 26, the quality rank has two levels: normal and degraded (or defective). The right vertical column is a determination condition, which is determined by the frame-by-frame quality determination unit 32 and output as “frame quality determination log”. Although the two-stage evaluation is normal and deteriorated (or defective) here, it is also possible to evaluate the deterioration step by step according to the number of frames in which “stranding” occurs.

The normal and deteriorated (or defective) or stepwise evaluation process has been described above. When the deterioration is determined, the display unit 34 notifies the maintenance worker that the handrail needs to be replaced. May be exchanged. In FIG. 26, a predetermined length of Sth is expressed by a frame, but it can also be expressed by a corresponding millimeter.
[Final decision means]
Next, the process of the final determination unit 33 will be described. The final determination unit 33 refers to the “good / bad determination log for each frame” and performs a process of making a quality determination as a single handrail to be inspected. For example, if there is no deterioration determination portion in the “good / good determination log for each frame”, the handrail can be determined as a non-defective product as a single unit. Further, if there is a portion where deterioration is determined more than a predetermined number of frames in the “good / bad determination log for each frame”, the handrail can be determined as a single unit for deterioration determination. Further, as a simple implementation example of the final determination unit 33, the worst quality determination result recorded in the “good / bad determination log for each frame” may be evaluated as a single unit of the handrail. The processing of the final determination unit 33 has been described above.
[Processing flow of image processing means]
FIG. 33 is a diagram showing a processing flow of the image processing means of the handrail inspection device for passenger conveyor of the present invention.

  With reference to FIG. 33, the processing flow of the image processing means of the handrail inspection device for the passenger conveyor will be described. Prior to processing a new handrail image, the memory and parameters used for processing are initialized (S50). Memory initialization includes initialization of “element feature log for each frame”, “trace log for each frame”, and “good / bad determination log for each frame”, for example.

  The frame processing loop S51 performs the loop processing until the moving image processing necessary for handrail quality evaluation is completed. When all the frames to be processed are completed, the process leaves the loop and proceeds to the processing of S58 (S52). The frame acquisition determination process S52 and the frame acquisition process S53 are performed by the frame acquisition unit 28. Next, SC detection processing S54 is performed in the SC detection unit 29. Then, the SC trace and element feature detection process S55 is performed in cooperation with the SC model holding unit 30 and the SC trace / element feature detection unit 31, and the SC model update process is simultaneously performed on the SC model held in the SC model holding unit 30. S56 is performed. Next, for each frame, a frame-by-frame quality determination process S57 is performed to update the “frame quality determination log”. This process is performed by the pass / fail judgment unit 32 for each frame.

After the above processing is completed for all necessary frames, the final determination unit 33 performs handrail quality final determination processing S58. At this time, the final determination result can be stored in a magnetic medium or transmitted to a quality control server or the like via a network. With the above processing, automatic determination of handrail quality evaluation is realized.
[Example of detecting the wire only in the part where the missing part occurs]
The processing of the strand detection processing flow (FIG. 1) in the strand detection unit 311 can be performed for all frames of the moving image. However, the element features detected by the SC trace / element feature detector 31 are performed in conjunction with the presence or absence of features other than “elementary wire”, ie, “contact”, “missing”, and “entanglement” features. You can also This is because “stranding” is likely to occur in a frame in which the characteristics of “contact”, “missing”, and “entanglement” are generated. In this way, by limiting the frames to be processed in the moving image, it is possible to reduce the processing amount of the entire moving image and contribute to the speeding up of the processing.
[Example of method in which luminance threshold Bth is not fixed]
The method of determining the luminance threshold value Bth for identifying whether the image is a penumbra that is a condition of the strand or an image having a main shadow that is a condition of the normal steel code has been described. The luminance values of the main shadow and penumbra change according to the output of the X-ray source, such as the dose related to the current and the wavelength related to the voltage, and also change depending on the type of scintillator used and the gain adjustment of the camera. . Therefore, in the above example, after setting the output of the X-ray source, the type of the scintillator, and the value of the camera gain, it is obtained by drawing as shown in FIGS. However, when the thickness of the rubber covering the steel cord varies greatly from place to place, the fixed luminance threshold Bth may not be sufficient.

  Therefore, when Bth is determined by the above drawing, KthB = Bth / Bbak, which is the ratio of this to the background luminance Bbak, is obtained. KthB is a rational number greater than 0 and less than 1. Then, when the position of the steel code that exists independently is determined by the SC trace / element feature detection unit 31, the luminance of the image between the steel codes can be measured, and this can be regarded as the background luminance Bbak. At this time, if the average value of the corresponding positions for several frames is taken as the background luminance Bbak, the influence of random noise can be offset. Then, using the assumed background luminance Bbak, the luminance threshold value Bth can be automatically adjusted according to changes in the situation such as rubber thickness, as Bth = KthB × Bbak.

  On the other hand, when Bth is determined by the above drawing, KthM = Bmin / Bth, which is a ratio with the main shadow luminance Bmin, can also be obtained. KthM is a rational number greater than 0 and less than 1. Then, when the position of the steel code that exists independently is determined by the SC trace / element feature detection unit 31, the luminance of the image of the steel code portion is measured, and this is regarded as the main shadow luminance Bmin. As described above, since the position of the steel code detected by the SC detection unit 29 is a valley of the luminance distribution, the image luminance at the position of the steel code can be regarded as the main shadow luminance Bmin. At this time, if the average value of corresponding positions for several frames is taken as the main shadow luminance Bmin, the influence of random noise can be offset. Then, using the assumed main shadow luminance Bmin, the luminance threshold value Bth can be automatically adjusted according to changes in the situation such as the rubber thickness, such as Bth = Bmin / KthM.

  The above measurement of Bbak and Bmin will be described with reference to FIG. FIG. 27 is a diagram for explaining a luminance measurement method when the luminance threshold value Bth is not fixed, and is a conceptual diagram of an X-ray image in which six normal steel cords exist. Unlike the other figures, the main shadow and penumbra are shown clearly. There are main shadows 87 to 92 at the center of each steel cord image, and there are penumbras 93 to 98 with hatched lines around them. An area 86 with a stippled portion of the upper half of the screen is an area that is darkly photographed as a whole because the rubber surrounding the steel cord is thick. If the luminance difference between the area 86 and other areas is large, it may be inappropriate to uniformly set the luminance threshold value Bth as a fixed value.

  Therefore, in the vicinity of each steel code, it is necessary to measure the luminance value of the main shadows 87 to 92 to be Bmin, and select the luminance of the region that is neither the main shadow nor the penumbra to be the background luminance Bbak. As described above, since the position of the steel cord detected by the SC detection unit 29 is a valley portion of the luminance distribution, the position of the steel code is in the range of the main shadows 87 to 92. Therefore, the main shadow luminance can be obtained by measuring the luminance at the position of the detected steel cord. On the other hand, the background luminance is the luminance at the midpoint position between the detected steel cords. When the steel cords are close to each other, the penumbra of adjacent steel cords may overlap and the background luminance may not be obtained. Therefore, when the penumbras of adjacent steel cords are close enough to overlap each other, it is necessary to stop the background luminance measurement and apply the background luminance value obtained before this. By the method described above, appropriate threshold luminance Bth and background luminance Bbak can be obtained even if the luminance varies depending on the rubber thickness. In the entire image, a corresponding region is provided around the detected steel code, and Bth and Bbak obtained with reference to the steel code position are applied in this region.

  In other words, the luminance range that is the criterion for penumbra detection by making a line is set with the background luminance of the projected image or the luminance of the main shadow as a parameter in consideration of the change in the luminance of the background of the projected image or the luminance of the main shadow. May be variably set.

Further, there is a boundary of a region 86 between the penumbra 95 and penumbra 96, and the background luminance between the steel cords related to the main shadow 89 and the main shadow 90 is averaged between the region 86 and the other regions. In some cases, the steel cord as a whole is a very narrow region, so this is not a problem.
[Example in which steel code tracing is omitted]
In the above embodiment (embodiment 1), it is judged whether it is good or bad by detecting all of “contact”, “disconnection”, “entanglement”, and “stretching” that can occur in a steel cord with a built-in handrail. I explained it on the assumption. However, when the quality determination of the steel cord (passenger conveyor) is performed only from the “stranding” feature, the quality determination can be performed more easily by executing the image processing of the second embodiment shown in FIG. . FIG. 28 is an image processing unit of the handrail inspection device for a passenger conveyor according to the second embodiment. The image processing unit 2 of the present embodiment includes a frame acquisition unit 28, a strand detection unit 311, a frame-by-frame quality determination unit 32, a final determination unit 33, a display unit 34, a command input unit 35, and a control unit 36. Consists of.

In the present embodiment, all frames of a moving image related to X-ray imaging are acquired by the frame acquisition unit 28, and the strand detection flow described with reference to FIG. . As a result, the presence / absence of “stranding” is determined for each frame, and the “element feature log for each frame” is output. This log records the presence / absence of “stranding” for each frame. The frame-by-frame quality determination unit 32, final determination unit 33, display unit 34, command input unit 35, and control unit 36 are the same as those in the first embodiment described with reference to FIG.
[Bidirectional trace method]
14 can hold the SC detection log table for each frame, and the SC model holding unit 30 can be configured to hold the SC model log table for each frame.

  The SC group model is used for collation with the detected steel code (SC) as described above for tracing the steel code. The SC model group is the SC data group (SC model) detected in the immediately preceding frame. Updated by log table). This is to improve accuracy by using the latest data of the steel code trace. FIG. 29 shows an example of the SC detection log table and the SC model log table. Here, in order to avoid the explanation complexity, a case where nine steel cords are designed is shown.

  For example, FIG. 29A shows an SC model log table in the 120th frame. The frame number 120 indicates that the frame corresponding to the log is 120, and the numbers 01 to 09 in the model code column indicate the numbers assigned to the model codes. On the other hand, FIG. 29B is an SC detection log table in the 121st frame corresponding to the next frame. The frame number 121 indicates that the frame corresponding to the log is 121, and the detection codes are arranged in ascending order of detected coordinates. “D” in the detection code column indicates that a steel code has been detected. The numbers 1 to 9 in the same column are the order of arrangement based on the coordinates, and are not the naming numbers of the detected codes. In this case, since all the steel cords in the design are detected, as described with reference to FIG. 20A, the detected coordinates are named 01 to 09 in ascending order of the coordinates of the detected steel cords. Thus, the SC model log table of the 121st frame corresponding to the next frame is obtained. This is shown in FIG.

  FIG. 30 shows another example of the SC detection log table and the SC model log table. Here, in order to avoid complication, a case where nine steel cords are designed is shown. For example, FIG. 30A is an SC model log table in the 120th frame. The frame number 120 indicates that the frame corresponding to the log is 120, and the numbers 01 to 09 in the model code column indicate the numbers assigned to the model codes. On the other hand, FIG. 30B is an SC detection log table in the 121st frame corresponding to the next frame. The frame number 121 indicates that the frame corresponding to the log is 121, and the detection codes are arranged in ascending order of detected coordinates. “D” in the detection code column indicates that a steel code has been detected. In this case, only seven steel cords are detected. When it is determined that, for example, 06 code and 07 code are not detected by the processing described with reference to FIG. 20B, 01 to 05 code, 08 code, and 09 code are detected as a steel code in 121 frames. Updated with coordinates. The undetected 06 code and 07 code are linearly interpolated between the detected 05 code coordinate and 08 code coordinate to update the 06 code and 07 code coordinates. With the above operation, even if there is an undetected steel code, the steel code model is updated.

  However, there is a problem that the update of the SC model becomes inaccurate if the number of undetected steal codes is large or a frame with undetected steal codes continues for a long time. Specifically, when the undetectable area of the steel code continues due to the entanglement of the steel code in the trace process in the forward direction, the range from the undetectable area to the length of a certain range (predetermined number of frames) is considered as an undetected area However, in reality, when tracing from the reverse direction, the steel code detectable area may be longer than the deemed undetected area. In order to solve such a problem of setting an undetected area that is not excessive, this embodiment also proposes a configuration in which the steel code is bidirectionally traced in the frame order and the frame reverse order. The use of bidirectional trace is optional.

  FIG. 31 is a diagram illustrating a method for realizing bidirectional tracing, and is a tracing method and a tracing apparatus configured by an SC detection unit 29, an SC model holding unit 30, and an SC trace / element feature detection unit 32. .

  The SC detection unit 29 has the SC detection log table 99 described with reference to FIGS. 29 and 30, and holds the coordinates of the steel code detected for each frame.

  The SC model holding unit 30 has the SC model log table forward direction 100 and the SC model log table reverse direction 101 described in FIG. 29 and FIG. 30, and sequentially updates the steel code model as described in FIG. A log of position coordinates together with the named code is held in the SC model log table forward direction 100.

  The SC trace / element feature detection unit 32 has an imperfect counter 102 and an imperfect counter log table 103. When receiving the coordinates of the steal code detected from the SC detection unit 29, the number of detected steal codes is determined. If it is less than the designed number, the imperfect counter 102 is incremented, and is reset when the number of detected steel cords matches the designed number.

  With this operation, the perfect counter 102 can know how many frames having undetected steal codes continue. The perfect counter log table 103 records and holds the value indicated by the perfect counter 102 for each frame. With this configuration, referring to the perfect counter log table 103 after the processing of all the frames is completed, it is possible to know the frame in which the steal code model has been updated in the forward trace in a state where the steal code has not been detected. it can.

  FIG. 32 is a diagram showing a processing flow of the image processing means of the handrail inspection device for the passenger conveyor of this embodiment when performing bidirectional tracing.

  With reference to FIG. 32, the processing flow of the image processing means of the handrail inspection device for the passenger conveyor will be described. Prior to processing a new handrail image, the memory and parameters used for processing are initialized (S50). Memory initialization includes, for example, “element feature log for each frame”, “trace log for each frame”, “pass / fail judgment log for each frame”, SC detection log table 99, SC model log table forward direction 100, There is initialization of SC model log table reverse direction 101, perfect counter 102 and perfect counter log table 103.

  The frame processing loop S51 performs the loop processing until the moving image processing necessary for handrail quality evaluation is completed. When all the frames to be processed are completed, the process leaves the loop and proceeds to the reverse loop processing of S60 (S52). The frame acquisition determination process S52 and the frame acquisition process S53 are performed by the frame acquisition unit 28. Next, SC detection processing S54 is performed in the SC detection unit 29. Then, the SC trace processing S59 is performed in cooperation with the SC model holding unit 30 and the SC trace / element feature detection unit 31, and at the same time, the SC model update processing S56 is performed for the SC model held in the SC model holding unit 30. . In the processes of S54, S59, and S56, the SC detection log table 99, the SC model log table forward direction 100, the perfect counter 102, and the perfect counter log table 103 are updated as described above. Also, element feature extraction and pass / fail judgment for each frame is performed during backward tracing, and thus is not performed here.

  The SC trace / element feature detection unit 32 resets the perfect counter after the end of the forward SC trace (S70), and then starts the SC trace in the reverse frame order (S60). At this time, from the SC detection unit 29, the number and coordinates of the steel codes obtained by referring to the SC detection log table 99 in the reverse order of the frames are used instead of the steel codes newly detected by the processing described with reference to FIGS. Obtain (S62).

  If the number of steel cords is less than the designed number, the perfect counter 102 is counted up, and if it matches the designed number, it is reset (S62).

  Further, the SC model is updated by the operation described in FIG. 20, and the SC model log table reverse direction 101 is updated (S63). At this time, the recording of the frame in the perfect counter 102 and the perfect counter log table 103 in the current frame is referred to. When the value recorded in the perfect counter log table 103 is smaller than the value recorded in the perfect counter log table 103 when the value of the perfect counter 102 is compared, the result of tracing in the forward direction first Can be judged to be more reliable. Therefore, in this case, the value of the SC model holding unit 30 is replaced with each steel code coordinate recorded in the SC model log table forward direction 100 (S63).

  By this process, the SC model can be updated with a more reliable value in the bidirectional trace. Using the results of SC trace, “contact”, “missing”, “entanglement”, “stranding” features are detected for each frame, and “element feature log for each frame” and “trace log for each frame” Is updated (S64). Next, for each frame, a frame-by-frame quality determination process S65 is performed, and the “frame-by-frame quality determination log” is updated. This process is performed by the pass / fail judgment unit 32 for each frame.

After the above processing is completed for all necessary frames, the final determination unit 33 performs handrail quality final determination processing S58. At this time, the final determination result can be stored in a magnetic medium or transmitted to a quality control server or the like via a network. With the above processing, automatic determination of handrail quality evaluation is realized.
[Wire detector using visible light]
Next, as another embodiment of the inspection device for the long member for the transfer mechanism, the strand detection of the wire or the rope strand not covered with the sheath is obtained by detecting the image with visible light instead of the X-ray. The device will be described.

  FIG. 34 is a diagram of an imaging system for detecting a strand with visible light. The cover 104 blocks external light, the camera 7 with a lens portion inserted into the cover 104, and a light source 107 having a finite size. And a screen 106 and an image processing unit 2 (not shown).

  The screen 106 is relatively thin like a shoji paper, and is a screen that can be photographed from the camera 7 side by forming a shadow (projected image) when an object is placed near the broken line 105. At this time, with respect to the object 109 having the reference outer diameter, the tangent line 110 is drawn from both ends of the light source 107 and the position 105 of the object is adjusted so as to intersect at one point on the screen 106. With this configuration, for the object 108 having an outer diameter larger than the reference outer diameter 109, a main shadow formed by completely shielding the light source 107 appears on the screen 106. On the other hand, for an object having an outer diameter smaller than the reference outer diameter 109, for example, a strand separated from a rope or wire, a penumbra formed by covering a part of the light source 107 appears. When this image is input to the image processing unit 2 via the camera 7 and the processing of the strand detection flow described with reference to FIG. 1 is performed, the strand can be detected.

  With this, the embodiment of the configuration of the image processing apparatus of the present invention has been described.

  According to the embodiment, when the occurrence of “stranding” of the steel cord in the longitudinal direction of the handrail continues for a predetermined length or longer, it is possible to determine that the quality of the handrail is defective, and thus the handrail needs to be replaced. It is possible to suppress misjudgment and unnecessary handrail replacement.

  In addition, when the “stranding” of the steel cord is not judged as defective, the quality of the handrail is determined to be degraded in consideration of the “stranding” of the steel cord. If it is determined that the quality of the handrail is deteriorated, it is known that the time for handrail replacement is close, and the quality of the handrail is divided into three or more stages (for example, defective, deteriorated, non-defective). Etc.) can be evaluated.

  The present invention has been described with reference to the embodiments. However, the configuration described in each of the embodiments so far is merely an example, and the present invention can be modified as appropriate without departing from the technical idea. .

  Further, the configurations described in the respective embodiments may be used in combination as long as they do not contradict each other. In the present specification, “inspection” may be referred to as “inspection”, and “inspection” in the present specification and claims includes “inspection”. Similarly, in this specification, “maintenance” may be referred to as “maintenance”, and “maintenance” in the present specification and claims includes “maintenance”. Maintenance (maintenance) means performing at least one of inspection (inspection), repair, and replacement, and includes a case where two or more of these are performed in combination.

DESCRIPTION OF SYMBOLS 1 ... X-ray imaging part 2 ... Image processing part 3 ... Encoder 4 ... Hand rail 4a ... Center position of the range in which a steel cord group exists generally 5 ... X-ray tube 6 ... Scintillator 7 ... Camera 8 ... Lead glass 9 ... Hand rail Crossing direction 10 ... Section 11 where the seam portion exists ... Elementary wire 12 ... X-ray exit 13 ... Range 14 irradiated with X-rays ... Boundary outer diameter 14a ... th outer diameter 15 ... Normal steel cord 16 ... Normal Steel cord 17 ... Two line segments 18 ... Two line segments 19 as tangent lines ... Two line segments 20 as tangent lines ... Penumbra area 21 ... Background area 22 ... Main shadow area 23 ... Image of normal steel cord Luminance distribution 24 of the shielded maximum penumbra area 25 ... penumbra area 26 ... luminance distribution 27 of the elemental image 27 ... luminance distribution 27α of the boundary outer diameter steel cord image ... luminance distribution 28 of the image of the outer diameter steel cord … Frame acquisition 29 ... SC detection unit 30 ... SC model holding unit 31 ... SC trace / element feature detection unit 311 ... strand detection unit 32 ... pass / fail judgment unit 33 ... final judgment unit 34 ... display unit 35 ... command input unit 36 ... control Section 37 ... Contrast correction curve 38 ... Positive threshold value 39 ... Negative threshold value 40 ... Luminance distribution valleys 41 to 42 ... Section 43 ... Maximum value 44 ... Minimum value 45 ... Steel cord 46 that exists independently ... Steel cord 47 that exists independently ... A state in which two steel cords are in contact 48 ... A short steel cord 49 ... A line that explicitly represents a luminance distribution analysis line 50 ... One of SC candidate cells 51 ... SC Result 52 of concatenating candidate cells ... Result of concatenating SC candidate cells 53 ... Result of concatenating SC candidate cells 54 ... Detected steel code 55 ... Detected steel code 56 ... Steel cord 57 detected ... Steel cord 58 detected ... Steel cord 59 detected ... Steel cord 60 detected ... Steel cord 61 detected ... Steel cord 62 detected ... Handrail X-ray image Schematic diagram 63 ... Detected steel code 64 ... Coordinate position 65 of SC model steel code ... Coordinate 66 where undetected steel code should be ... Schematic diagram of handrail X-ray image 67 ... Detected steel code 68 ... Steel model coordinate position 69 of SC model Steel cord 70 that has become thicker due to contact ... Conceptual diagram 71 of handrail X-ray image ... Steel code 72 detected ... Steel model coordinate position 73 of SC model ... Steel cord 74 that became thick due to the entanglement ... of normal steel cord A typical example of a normal steel code image 76 A typical example of a normal steel code image 77 A typical example of a normal steel code image 78 A typical example of a strand 79 to be detected Example of strand 80 ... Line segment total image 81 ... strand 82 ... strand 83 ... strand 84 ... strand 85 ... strand 86 ... dark area 87 because of rubber thickness ... normal steel cord main effect 88 ... Normal Steel Cord Shadow 89 ... Normal Steel Cord Shadow 90 ... Normal Steel Cord Shadow 91 ... Normal Steel Cord Shadow 92 ... Normal Steel Cord Shadow 93 ... Normal Steel Cord Shadow Penumbra 94 ... Normal steel cord penumbra 95 ... Normal steel cord penumbra 96 ... Normal steel cord penumbra 97 ... Normal steel cord penumbra 98 ... Normal steel cord penumbra 99 ... Steel Code detection log table 100 ... steel code model log table order 101 ... steel code model log table reverse 102 ... imperfect counter 103 ... imperfect counter log table 104 ... cover 105 for shielding external light ... object position 106 ... screen 107 ... light source 108 ... object cross section 109 having an outer diameter larger than the reference outer diameter ... object cross section 110 having a reference outer diameter ... drawing tangent

Claims (15)

An inspection device for a long member including a steel cord used for a transfer mechanism,
An imaging unit for irradiating X-rays or visible light to form a projection image of the steel cord;
When the projected image is input, and the input projected image includes a line segment that is thinner than a normal steel code and brighter than a normal steel code, image processing is performed to extract the line segment, An inspection device for a long member for a transfer mechanism, comprising: an image processing unit that detects stranding due to fraying of the steel cord to be inspected based on the line segment extraction data. In claim 1,
The photographing unit is set so as to form a main shadow for a steel cord having a normal outer diameter, and to form a penumbra narrower than a main shadow of a normal steel cord for stranding due to fraying of the steel cord. And
The image processing unit sets a luminance range which is a criterion for penumbra by the stranding, and inspects a long member for a transfer mechanism that extracts a line segment corresponding to the stranding based on the criterion. apparatus. In claim 1 or 2,
The image processing unit is an inspection device for a long member for a transfer mechanism that determines that the steel cord is made into a bare wire if the extracted line segment has a length equal to or greater than a determination reference value.   The long member for a transfer mechanism according to any one of claims 1 to 3, wherein the long member is a handrail for a passenger conveyor incorporating a steel cord, and the imaging processing unit is an X-ray imaging device. Inspection device.   The long member for a transfer mechanism according to any one of claims 1 to 3, wherein the long member is a bare steel cord serving as a wire rope for an elevator, and the photographing unit is a visible light projection imaging device. Inspection equipment.   6. The image processing unit according to claim 1, wherein the image processing unit traces a steel cord using a steel cord model for each captured image frame, and includes at least one of the disconnection, contact, and entanglement of the steel cord. An inspection device for a long member for a transfer mechanism that detects one.   7. The inspection device for a long member for a transfer mechanism according to claim 6, wherein the steel cord model is set to be updated by a detected steel cord photographed in the immediately preceding image frame.   8. The image processing unit according to claim 6, wherein the image processing unit performs the tracing of the steel code in both forward and reverse directions, and the backward tracing uses the steel code data stocked by the forward tracing. The inspection device for the long member for the transfer mechanism is set to be performed.   3. The apparatus for inspecting a long member for a transfer mechanism according to claim 2, wherein the luminance range that is a criterion for determining a penumbra by forming a strand is variable with the luminance of the background of the projected image or the luminance of the main shadow as a parameter. In any one of Claims 1 thru | or 9,
The imaging unit may have an X-ray tube having an X-ray exit having a larger vertical and horizontal size than the outer diameter of the steel cord that is to be detected or the outer diameter of the steel cord that is to be detected. A steel cord having a visible light source having a visible light exit having a large vertical and horizontal size, and having an outer diameter intermediate between the outer diameter of the wire steel cord to be detected and the outer diameter of a normal steel cord A transfer mechanism in which the X-ray tube or visible light source, a long member including a steel cord for inspection, and the conversion plate are arranged in a geometrical relationship in which the main shadow is not projected onto the conversion plate that projects a projection image. Long member inspection device. In any one of Claims 1 thru | or 10,
The image processing unit has a steel code model that can update and maintain the number of steel codes in design and the center of gravity coordinates of each steel code, and each detected steel code is compared with each steel code center of gravity coordinate of the steel code model. An inspection device for a long member for a transfer mechanism that updates each barycentric coordinate of the steel cord model so that the sum of differences is minimized. In claim 11,
In the image processing unit, for each frame, the center of gravity coordinates of the detected steel cord, the updated center of gravity coordinates of the steel cord model, the number of detected steel cords, and the number of detected steel cords are designed. Inspection device for a long member for a transfer mechanism that retains in a memory the value of a frame counter that counts the number of continuing frames of a frame that is less than the number of steel codes in the memory as log data for each frame. In the passenger rail handrail inspection device according to claim 12,
The image processing unit has a second steel code model, and after updating the steel code model, tracing is completed in the forward direction for the entire captured video, and then in descending order from the largest frame number for each frame. The barycentric coordinates of the second steel cord model so that the sum of the differences between the detected steel cords held in the log data and the barycentric coordinates of the steel cords of the second steel cord model is minimized. Trace and update in the opposite direction,
In the frame, when the value of the log data of the frame counter in the frame when traced in the forward direction is smaller than the value of the frame counter when traced in the reverse direction, the second steel An inspection device for a long member for a transfer mechanism that replaces each barycentric coordinate of the code model with the coordinate of the steel code model when traced in the forward direction. A method for inspecting a long member including a steel cord used for a transfer mechanism,
Irradiate X-rays or visible light to form a projection image of the steel cord,
When the projected image is input, and the input projected image includes a line segment that is thinner than a normal steel code and brighter than a normal steel code, image processing is performed to extract the line segment, An inspection method for a long member for a transfer mechanism, characterized in that a stranding due to fraying of the steel cord to be inspected is detected based on the line segment extraction data. In claim 14,
In the step of photographing the projection image, a main shadow is formed for a steel cord having a normal outer diameter, and a penumbra thinner than a main shadow of a normal steel cord is formed for stranding by fraying of the steel cord. Is set to
In the image processing step, a long range member for a transfer mechanism that sets a luminance range that is a determination criterion for penumbra by the formation of a strand and extracts a line segment corresponding to the formation of a strand based on the determination criterion. Inspection method.

Images