JPWO2021014645A1 - Inspection equipment and methods, as well as programs and recording media - Google Patents

Abstract

The sensor (25) detects the appearance or surface shape of the linear portion extending in the second direction at different positions in the first direction to be inspected. The processing circuit (32) acquires surface data representing the appearance or surface shape at different positions in the first direction from the detection result of the sensor (25), and obtains the surface data and the first reference data (Ra) and the first reference data (Ra). The first and second similarity with the reference data (Rb) of No. 2 are calculated, and the coordinates having the first and second similarity (p, q) as elements are calculated. A group of coordinates is generated by repeating the same process as above for a plurality of positions in the first direction of the inspection target or an object having the same appearance or surface shape as the inspection target. The degree of deviation of the coordinates of the position of interest from the group of the above coordinates is calculated as an isolation index, and the presence or absence of an abnormality is determined based on the isolation index. It is possible to reduce the error in determining the presence or absence of an abnormality.

Description

The present invention relates to an inspection device and a method. The present invention particularly relates to an inspection device and a method for detecting an inspection target and detecting an abnormality in the inspection target from surface data obtained as a result of the detection. The present invention also relates to programs and recording media.

Patent Document 1 describes an apparatus for inspecting a long body having a periodic pattern. The apparatus described in Patent Document 1 collects surface data of a long body moving in a long direction, calculates the degree of similarity between the surface data and the first reference data and the second reference data, and calculates the similarity. Anomalies are detected based on the locus of a similarity vector whose elements are degrees on a two-dimensional plane. For example, anomalies are detected based on the frequency with which the norm of the similarity vector falls below the threshold.

International Publication No. 2016/157290

In the apparatus described in Patent Document 1, when setting a threshold value for a norm, it is necessary to take a certain margin, and therefore, the presence or absence of an abnormality may be erroneously determined.

An object of the present invention is to reduce an error in determining the presence or absence of an abnormality in an inspection target.

The inspection device of the present invention is
A sensor that detects the appearance or surface shape of a linear portion extending in a second direction different from the first direction at each of different positions in the first direction to be inspected.
One of the different positions is selected as the position of interest, and surface data consisting of a sequence of detection values representing the appearance or surface shape of the linear portion of the position of interest is acquired from the detection result of the sensor. The first similarity, which is the similarity between the surface data and the first reference data, and the second similarity, which is the similarity between the surface data and the second reference data, are calculated, and the first similarity is calculated. And the coordinates having the second similarity as an element, or the coordinates having two functions having the first similarity and the second similarity as variables are calculated.
The same processing as the acquisition of the surface data for the linear portion of the attention position, the calculation of the first similarity and the second similarity, and the calculation of the coordinates is performed on the inspection target or the same appearance as the inspection target. Alternatively, the attention to the group of coordinates obtained by performing on the linear portion extending in the second direction at a plurality of positions in the first direction of the product or work piece having a surface shape or the portion thereof. The degree of deviation of the coordinates calculated for the linear part of the position is calculated as an isolation index.
It has a processing circuit for determining whether or not there is an abnormality in the inspection target based on the isolation index.

The inspection method of the present invention is
At each of the different positions in the first direction of the inspection target, the appearance or surface shape of the linear portion extending in the second direction different from the first direction is detected.
One of the different positions is selected as the position of interest, and from the result of the detection, surface data consisting of a sequence of detection values representing the appearance or surface shape of the linear portion of the position of interest is acquired, and the surface data is obtained. The first similarity, which is the similarity between the data and the first reference data, and the second similarity, which is the similarity between the surface data and the second reference data, are calculated, and the first similarity and the first similarity are calculated. Calculate the coordinates having the second similarity as an element, or the coordinates having two functions having the first similarity and the second similarity as variables as elements.
The same processing as the acquisition of the surface data for the linear portion of the attention position, the calculation of the first similarity and the second similarity, and the calculation of the coordinates is performed on the inspection target or the same appearance as the inspection target. Alternatively, the attention to the group of coordinates obtained by performing on the linear portion extending in the second direction at a plurality of positions in the first direction of the product or work piece having a surface shape or the portion thereof. The degree of deviation of the coordinates calculated for the linear part of the position is calculated as an isolation index.
Based on the isolation index, it is determined whether or not the inspection target has an abnormality.

According to the present invention, it is possible to reduce an error in determining the presence or absence of an abnormality in the inspection target.

It is a figure which shows the steel cable which is an example of an inspection target. (A) is a schematic elevation view showing a camera and a moving mechanism used in the inspection device of the first embodiment of the present invention, and (b) is a downward view from the position of line 2B-2B in FIG. 2 (a). The seen schematic plan view (c) is a schematic plan view seen downward from the position of line 2C-2C in FIG. 2 (a). It is a figure which shows the positional relationship between a steel cable and one camera. It is a functional block diagram which shows the structure of the inspection apparatus of Embodiment 1. FIG. It is a functional block diagram which shows the structure of the processing circuit of FIG. It is a functional block diagram which shows the structure of the 1st abnormality detection part of FIG. It is a figure which superimposes and shows the example of the image obtained by the image pickup on the steel cable which is the image pickup target. It is a figure which shows an example of the surface data obtained from one row of FIG. 7. It is a waveform diagram which shows the 1st and 2nd reference data used in the abnormality detection part of FIG. It is a figure which shows an example of the locus of a point represented by the coordinates calculated based on the surface data. It is a figure which shows the other example of the locus of a point represented by the coordinates calculated based on the surface data. An example of the positional relationship between a point represented by coordinates calculated based on surface data for a position of interest and a point represented by a large number of coordinates calculated based on surface data for positions other than the position of interest. It is a figure which shows. It is a flowchart which shows the procedure of calculation of the isolation index in Embodiment 3 of this invention. In the third embodiment, a point represented by coordinates calculated based on surface data for a position of interest and a point represented by a large number of coordinates calculated based on surface data for positions other than the position of interest. It is a figure which shows an example of the positional relationship of. It is a functional block diagram which shows the abnormality detection part used in the inspection apparatus of Embodiment 4 of this invention. It is a waveform diagram which shows the 1st and 2nd reference data generated by the abnormality detection part of FIG. It is a figure which shows an example of the distribution of the point represented by the coordinates calculated based on the surface data in the abnormality detection part of FIG. It is a block diagram which shows the abnormality detection part used in the inspection apparatus of Embodiment 5 of this invention. It is a functional block diagram which shows the abnormality detection part used in the inspection apparatus of Embodiment 6 of this invention. FIG. 5 is a diagram showing an example of an image obtained by imaging in the inspection device of the sixth embodiment superimposed on a steel cable to be imaged. It is a waveform diagram which shows the example of the data obtained from the different rows of FIG. It is a figure which shows an example of the distribution of the point represented by the coordinate calculated based on the surface data in the abnormality detection part of FIG. In the anomaly detection unit of FIG. 19, it is a figure which shows an example of the positional relationship between the point represented by the coordinates calculated based on the surface data about the attention position, and the point represented by a large number of coordinates prepared in advance. be. In the abnormality detection unit of FIG. 19, when bulk processing is performed, it is calculated based on the points represented by the coordinates calculated based on the surface data for the position of interest and the surface data for the positions other than the position of interest. It is a figure which shows an example of the positional relationship with the point represented by a large number of coordinates. It is a functional block diagram which shows the abnormality detection part used in the inspection apparatus of Embodiment 7 of this invention. It is a figure which shows the other example of the image which can be the object of the inspection apparatus of this invention. It is a waveform diagram which shows an example of the surface data obtained from one row of the image of FIG. 26. It is a waveform diagram which shows the example of the 1st and 2nd reference data used in the calculation of the similarity with respect to the surface data shown in FIG. 27. It is a figure which shows an example of the distribution of the point represented by the coordinates calculated based on the surface data obtained from the image of FIG. 26. It is a figure which shows the other example of the image which can be the object of the inspection apparatus of this invention. It is a figure which shows an example of the distribution of the point represented by the coordinate calculated based on the surface data obtained from the image of FIG. (A) is a diagram showing another example of an image that can be a target of the inspection device of the present invention, and (b) is a waveform diagram showing a change in brightness of the image of FIG. 32 (a). It is a waveform diagram which shows an example of the surface data obtained from one row of the image of FIG. 32 (a). It is a waveform diagram which shows the example of the 1st and 2nd reference data used in the calculation of the similarity with respect to the surface data shown in FIG. 33. It is a figure which shows an example of the distribution of the point represented by the coordinates calculated based on the surface data obtained from the image of FIG. 32 (a). It is a figure which shows the other example of the image which can be the object of the inspection apparatus of this invention. (A) is a waveform diagram showing an example of surface data obtained from one row of the image of FIG. 36, and (b) is a first used for calculating the similarity of surface data shown in FIG. 37 (a). It is a waveform diagram which shows the example of the 2nd reference data. It is a figure which shows an example of the distribution of the point represented by the coordinates calculated based on the surface data obtained from the image of FIG. 36. It is a figure which shows the other example of the image which can be the object of the inspection apparatus of this invention.

Embodiments of the present invention will be described with reference to the accompanying drawings. In the attached figure, the same reference numeral indicates the same part or the corresponding part.

Embodiment 1.
The inspection device of the first embodiment takes an image of the peripheral surface of the steel rope with a camera while moving the steel rope in the long direction of the steel rope as an inspection target, and from the image data obtained by the imaging, the presence or absence of an abnormality in the appearance. Is to inspect.

The steel rope has multiple strands. For example, FIG. 1 shows a steel cable 11 having eight strands 12.

As shown in the figure, the steel cable 11 is formed by twisting strands 12. Each strand 12 is spirally wound. As a result, irregularities corresponding to the thickness of each strand are regularly arranged on the peripheral surface of the steel cable 11.

The position of the unevenness in the circumferential direction of the steel rope 11 changes according to the position of the steel rope in the long direction, and is the same for each strand pitch Ps. The twist pitch Pt is equal to the product of the strand pitch Ps and the number of strands 12.

Examples of the steel rope 11 include a steel rope for ships, a steel rope for fishing, a steel rope for cranes, a steel rope for civil engineering work, a steel rope for cable cars, a steel rope for elevators, and a steel rope for bridges.

FIG. 2A is a schematic elevational view showing the camera and the moving mechanism 20 used in the inspection device 30 of the first embodiment. FIG. 2B is a schematic plan view looking downward from the position of line 2B-2B in FIG. 2A. FIG. 2C is a schematic plan view looking downward from the position of line 2C-2C in FIG. 2A.

The moving mechanism 20 has a roller group 22 rotatably held by the frame 21.
The roller group 22 includes the first set of rollers 22-11 to 22-14 and the second set of rollers 22-21 to 22-24 arranged at different positions of the MV in the moving direction (vertical direction in the figure) of the steel rope 11. , A third set of rollers 22-31 to 22-34, and a fourth set of rollers 22-41 to 22-44.

As shown in FIG. 2B, the first set of rollers 22-11 to 22-14 are provided so as to sandwich the steel rope 11 from four directions. The rollers of the second set, the third set, and the fourth set are also provided so as to sandwich the steel rope 11.

The first to fourth sets of rollers are driven by a drive mechanism (not shown) to rotate, and the friction between the surface of each roller and the surface of the steel rope 11 causes the steel rope 11 to move in its long direction. For example, in FIG. 2A, it is moved upward.
Instead of driving all the rollers of the 1st to 4th sets by the drive mechanism, only a part of the rollers of the 1st to 4th sets is driven by the drive mechanism, and the rest are the steel rope 11. You may try to take it around by the friction of.

The inspection device 30 includes cameras 25-1 to 25-4 as sensors. As shown in FIG. 2C, the cameras 25-1 to 25-4 are arranged so as to surround the steel rope 11, and the steel rope 11 is imaged from four directions to obtain an image of the entire circumference of the steel rope 11. ..
In the illustrated example, the steel rope 11 is imaged from four directions, but instead, the steel rope 11 may be imaged from three directions to obtain an image of the entire circumference of the steel rope 11.
A light source (not shown) for illuminating the portion of the steel cable 11 to be imaged may be provided.

The cameras 25-1 to 25-4 are fixed to the frame 21, and the distance between the cameras 25-1 to 25-4 and the peripheral surface of the steel cable 11 which is the surface to be inspected is kept constant.

In FIG. 2A, the rollers 22-14, 22-24, 22-34, and 22-44 in front of the steel cable 11, the cameras 25-4, and the rollers and cameras of the frame 21 are shown. The illustration of the supporting part is omitted.

In the inspection device of the present embodiment, the moving mechanism 20 moves the steel rope 11 in its long direction to change the relative position of the steel rope 11 with respect to the cameras 25-1 to 25-4, while changing the relative position of the steel rope 11 with respect to the cameras 25-1 to 25-1. According to 25-4, the peripheral surfaces of the steel rope 11 are sequentially imaged at different positions in the long direction of the steel rope 11, and the abnormality of the steel rope 11 is detected based on the image pickup data.

FIG. 3 shows the positional relationship between the steel cable 11 to be inspected and one camera 25-1. xyz is an axis for expressing the coordinates in the three-dimensional space, and the long DL in the steel cable 11 is defined as the y-axis, and the x-axis and the z-axis are orthogonal to this. In the illustrated example, the steel rope 11 moves in the positive direction of the y-axis.
The positional relationship between the other cameras 25-2 to 25-4 and the steel cable 11 is the same as described above.

Each of the cameras 25-1 to 25-4 is arranged so as to take an image of a portion (linear portion) along a line extending in the circumferential direction of the steel cable 11. The circumferential direction of the steel rope is a direction perpendicular to the moving direction of the steel rope when viewed from each of the cameras 25-1 to 25-4. Therefore, each camera is arranged so as to capture a linear portion extending in a direction perpendicular to the moving direction of the steel cable when viewed from the camera.

Each of the cameras 25-1 to 25-4 may be an area camera or a line camera.
When each of the cameras 25-1 to 25-4 is a line camera, each camera is arranged so that the direction in which the image pickup elements are arranged corresponds to the direction perpendicular to the moving direction MV of the steel cable 11. , The linear portion extending in the circumferential direction of the steel cable 11 is imaged. The linear portion of the steel cable 11 is imaged by all or part of the image pickup element of the line camera. In order to increase the resolution, it is desirable to take images with as many image sensors as possible.

When each of the cameras 25-1 to 25-4 is an area camera, each camera has a direction in which the direction of the line (horizontal line) of the imaging screen corresponds to the direction perpendicular to the moving direction MV of the steel cable. The linear portion extending in the circumferential direction of the steel rope 11 is imaged by the image pickup elements arranged in the row direction. The linear portion of the steel cable 11 is imaged by all or part of the image sensor in one row of the area camera. In order to increase the resolution, it is desirable to take images with as many image sensors as possible.

The imaging range of each camera in the circumferential direction of the steel cable 11 is slightly larger than 1/4 of the entire circumference of the steel cable 11, and the above imaging ranges overlap each other at their ends between adjacent cameras. As a result, any position in the circumferential direction of the steel cord is included in at least one imaging range of the four cameras.

FIG. 4 is a diagram showing the configuration of the inspection device 30 of the first embodiment.
The illustrated inspection device 30 includes cameras 25-1 to 25-4 shown in FIGS. 2A and 2C, and a processing circuit 32.

The processing circuit 32 may be configured by dedicated hardware, or may be configured by a processor and a memory.

When configured with dedicated hardware, the processing circuit may be, for example, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

When composed of a processor and a memory, the processing circuit may be realized by software, firmware, or a combination of software and firmware. Software or firmware is written as a program and stored in memory. The processor realizes the function of each processing circuit by reading and executing the program stored in the memory.

Here, the processor may be, for example, a CPU (Central Processing Unit), an arithmetic unit, a microprocessor, a microcomputer, or a DSP (Digital Signal Processor).
The memory may be, for example, a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Project ROM), or an EPROM (Electrically EPROM). , A magnetic disk such as a hard disk, or an optical disk such as a CD (Compact Disc) or a DVD (Digital Versaille Disc).

The processing circuit 32 may be partially realized by dedicated hardware and the other part may be realized by software or firmware.

FIG. 4 shows a configuration in which the processing circuit 32 is realized by a computer 34 including one processor.
The illustrated computer 34 includes a processor 341 and a memory 342.
The memory 342 stores a program executed by the processor 341.
The processor 341 reads the program stored in the memory 342 and executes it.

The image pickup data from the cameras 25-1 to 25-4 are taken into the computer 34 via an interface (not shown).

FIG. 5 is a functional block diagram showing the configuration of the processing circuit 32 of FIG.
As shown in FIG. 5, the processing circuit 32 includes first to fourth abnormality detection units 35-1 to 3-5-4 and an inspection result output unit 36.
The first to fourth abnormality detection units 35-1 to 3-5-4 are provided corresponding to the cameras 25-1 to 25-4, respectively, and receive image pickup data from the corresponding cameras to correspond to each other. Detects abnormalities in the imaging range of the camera.

The first to fourth abnormality detection units 35-1 to 3-5-4 have the same configuration as each other.
Hereinafter, the first abnormality detection unit 35-1 will be described.

FIG. 6 shows the configuration of the first abnormality detection unit 35-1.
The abnormality detection unit 35-1 shown in FIG. 6 includes a data acquisition unit 41, a reference data storage unit 42, a similarity calculation unit 43, a coordinate calculation unit 45, a coordinate storage unit 46, and an isolation index calculation unit. 47 and an abnormality determination unit 48 are provided.

The data acquisition unit 41 receives the image pickup data from the camera 25-1 and generates surface data Dk for each of the linear portions at different positions in the elongated direction of the steel cable 11. The position in the long direction is sometimes called the long direction position. The surface data for the linear portion of each long direction position may be referred to as the surface data for the long direction position.

If the camera 25-1 is a line camera, the portion corresponding to the captured linear portion of the one-dimensional image obtained by imaging at each long direction position is the surface data Dk for the linear portion. Become.
If the camera 25-1 is an area camera, the data of the portion of the two-dimensional image obtained by imaging at each long direction position that belongs to a predetermined row and corresponds to the captured linear portion is , The surface data Dk for the linear portion.

In generating the surface data Dk, bias removal, noise removal, or the like may be performed on the imaging data.
Bias removal is a process of removing low frequency components in the imaging data. The low frequency component referred to here is a component of a change in brightness due to the fact that the steel rope is columnar as a whole, that is, the envelope surface of the peripheral surface of the steel rope is columnar. That is, in each camera, the brightness of each part of the peripheral surface of the steel cable when viewed from the camera changes depending on the angle of the part with respect to the optical axis of the camera. The angle of each portion of the peripheral surface of the steel rope with respect to the optical axis of the camera is determined by the angle of the envelope surface of the peripheral surface of the steel rope with respect to the optical axis of the camera and the unevenness due to the strand 12. Bias removal is performed to remove the influence of the change in the angle of the envelope surface with respect to the optical axis of the camera. By removing the bias, it is possible to extract only the component of the change in brightness due to the unevenness of the strand 12.

Noise removal is a process for removing high frequency noise in an image. The high frequency referred to here means a frequency higher than the frequency due to the unevenness of the strand 12 and the frequency due to the abnormality to be detected.

The surface data Dk belongs to a portion of the one-dimensional image data corresponding to the linear portion or one row of the two-dimensional image, and the pixel values Dk ₁ , k of a fixed number of pixels in the portion corresponding to the linear portion. ₂ , ... Dk _J are arranged in the order of pixels, and can be seen as a pixel value vector.
The pixel value referred to here may be a luminance value or a color component value. When it is a color component value, it may be, for example, one of R, G, and B. In the following, it is assumed that the luminance value is used, and the magnitude of the pixel value represents the brightness.
Since the pixel value constituting the surface data Dk is a value obtained by detection by a camera as a sensor, it may be referred to as a detected value.

The pixel values constituting the surface data Dk for each long direction position change according to the unevenness of the peripheral surface of the steel cable 11 due to the strand 12. That is, the surface data Dk has periodicity corresponding to the period of unevenness in the circumferential direction of the peripheral surface due to the strand 12.
Since the image obtained by each camera is a projection of the peripheral surface of the steel cord on a flat imaging screen, the length in the circumferential direction does not completely correspond to the length in the captured image, and the steel cord does not completely correspond to the length in the captured image. Even if the cycle on the peripheral surface is constant, the length of one cycle changes depending on the row direction position in the captured image, and the change in pixel value (change with respect to the row direction position) also in the surface data Dk. The cycle (spatial cycle in the image) changes. However, by limiting the imaging range, especially the range in which surface data is acquired, to the portion near the center in the width direction of the steel cable when viewed from the camera (excluding the portion near the edges on both sides), the influence of the above changes can be reduced. It can be small and negligible.

When the position in the long direction imaged by the movement of the steel cable 11 changes, the pixel value vector composed of the surface data Dk also changes. This change has periodicity, and the period (spatial period) is equal to the strand pitch Ps.

FIG. 7 shows, when the camera 25-1 is an area camera, an example of an image obtained by imaging with the camera is superimposed on the steel cable 11 to be imaged. FIG. 7 shows the range of the steel rope 11 to be imaged by the dotted frame 14, and shows the outline of the image Ga obtained by the image inside the frame 14.
In the image Ga obtained by imaging, Gaa near the center of each strand 12 is the brightest, and Gab near the boundary between adjacent strands 12 is the darkest. In the actual image, the brightness of each part of the image gradually changes continuously, but in FIG. 7, the bright part Gaa is shown by white and the dark part Gab is shown by a dot pattern for simplification.

In the present embodiment, from the image Ga within the range shown by the dotted frame 14, the data of the row Lk specified in advance is extracted and used as the surface data Dk.
FIG. 8 shows an example of surface data Dk obtained from row Lk of FIG. In FIG. 8, the vertical axis indicates the brightness of the image, and the horizontal axis indicates the row direction position in the frame 14 of the dotted line in FIG. 7. _{The x l} and x _{r in} FIG. 8 correspond to _{the left end x l} and the right end x _r of the dotted frame 14 in FIG. 7.
In the example shown in FIG. 8, the pixel value changes with the period Tk.

When the peripheral surface of the steel cable 11 is imaged while the steel cable 11 is moved in the long direction, the surface data Dk (pixel value vector) obtained by the imaging changes continuously and periodically with respect to the moving distance. The periodic change is caused by the steel rope 11 having periodicity in the longitudinal direction. The change is continuous because the strand 12 (thus, the bright and dark parts due to the unevenness of the strand 12) extends diagonally in the longitudinal direction, that is, in the moving direction, so that the pixel value vector gradually becomes. Because it changes.

The reference data storage unit 42 stores the first reference data Ra and the second reference data Rb. The first reference data Ra and the second reference data Rb are, for example, prepared in advance. The first reference data Ra and the second reference data Rb in the illustrated example have a period equal to the period Tk of the surface data described above.

For example, if the period Tk of the change in the pixel value (change with respect to the row direction position) constituting the surface data Dk obtained when there is no abnormality in the steel cable 11 is known, the period Tk is obtained and the phase difference between the two is known. A vector consisting of a sequence of values representing two sine waves of π / 2 can be used as the first reference data Ra and the second reference data Rb.
FIG. 9 shows an example of such sinusoidal reference data Ra, Rb. In FIG. 9, Tk represents a period.

Each of the first and second reference data Ra and Rb is a vector having the same length as the surface data Dk (data consisting of a sequence of values having the same number as the number of pixel values constituting the surface data Dk). Therefore, the first reference data Ra is _{represented by the values Ra 1} , Ra ₂ , ... Ra _J , and the second reference data Rb is represented by the values Rb ₁ , Rb ₂ , ... Rb _J.
Each of the first and second reference data Ra and Rb has a wave number equal to or close to the wave number of the change in the brightness of the pixel values constituting the surface data Dk.

The period of the sine wave represented by each of the first and second reference data Ra and Rb does not have to be exactly the same as the period Tk of the surface data Dk, and in short, it may have the same or approximate period.

Since the sine waves represented by the first and second reference data Ra and Rb have a phase difference of π / 2, the first reference data Ra and the second reference data Rb are orthogonal to each other. ..
As described above, when the first reference data Ra is _{represented by (Ra 1} , Ra ₂ , ... Ra _J ) and the second reference data Rb is represented by (Rb ₁ , Rb ₂ , ... Rb _J ). , The fact that the first reference data Ra and the second reference data Rb are orthogonal to each other means that the following equation (1) is satisfied.

The two sine waves represented by the first and second reference data Ra and Rb do not have to have a phase difference of exactly π / 2, and may have a value close to π / 2. That is, the first reference data Ra and the second reference data Rb do not have to be exactly orthogonal to each other, and may have an approximate relationship with the orthogonal relationship.

The similarity calculation unit 43 calculates the first similarity p, which is the similarity between the surface data Dk and the first reference data Ra for each long direction position output from the data acquisition unit 41.
The similarity calculation unit 43 also calculates a second similarity q, which is the similarity between the surface data Dk and the second reference data Rb.

式（２ａ）及び（２ｂ）で、Ｅ［−］は期待値を表す。 _{In this embodiment, the correlation coefficients ρ a} and ρ _b represented by the following equations (2a) and (2b) are used as the similarity p and q.
The correlation coefficients ρ _a and ρ _b are defined as follows.

In the equations (2a) and (2b), E [−] represents the expected value.

The coordinate calculation unit 45 obtains two-dimensional coordinates of an orthogonal coordinate system having a first degree of similarity p and a second degree of similarity q as elements. Such two-dimensional coordinate represented by _S i (p, q). S _i (p, q) may be simply represented by _{S i.} Further, the points at the positions on the two-dimensional plane represented by the two- _{dimensional coordinates Si are also represented by the same reference numerals S i} (p, q) or S _i .
An example of such a point S _i (p, q) is shown in FIG. When representing the degree of similarity in the correlation coefficient of the above formula (2a) and (2b), the distance from the origin O to the point S _i, i.e. norm is the value less than 1.

The process from the calculation of the similarity to the calculation of the coordinates is performed for each position in the length direction, and one point _Si is obtained for each position in the length direction (hence, for each linear portion).

As described above, the pixel value vector composed of the surface data Dk obtained by imaging the peripheral surface of the steel rope 11 while moving the steel rope 11 in the long direction is continuously and with respect to the moving distance. It changes periodically. Therefore, point S _i is moved on a two-dimensional plane so as to draw a predetermined trajectory. This locus is, for example, a circle centered on the origin O.

The direction of movement (rotation) along the circle of the point S _i is dependent on the relative movement direction of the steel ropes with respect to the camera 25-1. The period required for one rotation corresponds to one strand pitch Ps of the steel rope. That is, when the steel cable moves by one strand pitch Ps with respect to the camera, the point _Si makes one rotation and returns to the original position.

Thickness and strand pitch Ps of the strands 12 is constant, if there is no abnormality in the wire rope, the point S _i draw repeating the same trajectory regardless of the speed of movement.

Although locus point S _i is drawn in the example shown in FIG. 10 is a circular shape, there not necessarily to be a circular shape, may comprise, for example, an elliptical shape as shown in FIG. 11.

If there is an abnormality on the peripheral surface of the steel cable 11, the surface data obtained by imaging the portion with the abnormality will be different from the case where there is no abnormality, and the point _Si will be located at a position deviating from the above trajectory. appear.

The isolation index calculation unit 47 calculates the degree of deviation, that is, the degree of deviation.
In the following, the processing target, that is, the position in the long direction that is the target of calculation of the isolation index is referred to as the attention position, and the coordinates obtained from the surface data for the attention position are simply "coordinates calculated for the attention position". Alternatively, it may be referred to as "coordinates about the position of interest".

In the part of the inspection device 30 composed of the camera 25-1 and the first abnormality detection unit 35-1, the camera 25-1 and the data acquisition unit 41 position the steel rope 11 at a plurality of different positions in the elongated direction. for each, acquires surface data, performs coordinate calculation of by the similarity calculation and coordinate calculation unit 45 of by the similarity calculation unit 43 based on the surface data, the coordinate storage unit 46, storing the coordinates S _i calculated doing, form a group consisting of a plurality of coordinates S _i, in solitary index calculation unit 47, for the group of forming coordinate S _i, solitary index deviance coordinates S _i calculated for the target position If the isolation index is larger than the threshold value, the abnormality determination unit 48 determines that there is an abnormality.

Hereinafter, refers to a group consisting of a plurality of coordinates and coordinate group, they say the point group of the group of points S _i of a plurality of coordinates.
The number of coordinates that make up the coordinate group, and therefore the number of points that make up the point cloud, is set large enough so that it is not affected by fluctuations in the surface data, for example, a number larger than a predetermined number. There is.
As described above, the above numbers are large enough to be referred to as "many" below.

All coordinates belonging to the coordinate group, hence S _i points belonging to the point group is desirably those obtained when there is no abnormality in steel ropes. However, it may include points obtained when there is an abnormality. This is because when calculating the isolation index, comparisons are made with a large number of coordinates, that is, a sufficiently large number of coordinates.

Coordinate storage unit 46 stores a number of coordinates belonging to coordinate group described above, i.e. the coordinates for many points S _i belonging to the point group of the above.
A large number of coordinates belonging to a coordinate group are, for example, a large number of different parts of the steel rope 11 to be inspected before the inspection is performed, or a large number of steel ropes (indicated by the same reference numeral 11) different from the steel rope 11 to be inspected. The coordinates may be obtained by acquiring surface data by performing imaging at the position in the long direction of the above, and calculating the similarity and the coordinates with respect to the acquired surface data.
The above "other wire rope" needs to have the same appearance or surface shape as the wire rope to be inspected, for example, the same specifications as the wire rope to be inspected. This point is the same in the following other embodiments.

It represents a point corresponding to coordinates calculated from surface data obtained before the implementation of the test in Sa _i, represents a group of points consisting of point Sa _i in GSa.

In this case, the inspection device 30 operates in the point cloud generation mode prior to the operation mode for performing the inspection (inspection execution mode).
In the point cloud generation mode, the peripheral surface of the steel rope 11 is imaged while moving the steel rope or another steel rope 11 to be inspected, so that surface data Dk is sequentially acquired for a large number of long direction positions, and each surface is sequentially acquired. reference data Ra, similarity p and Rb, the q determined for data Dk, the similarity p, determine coordinates _Sa i (p, q) from q.
In this case, it is desirable that the portion of the steel cable 11 to be imaged is a portion having no abnormality.

Obtaining surface data, the calculation and the coordinate calculation of similarity, different number of (N number of) multiple points Sa i by performing the longitudinal direction position of steel ropes _{11 (i = 1,2, ...,} N ) Is formed. N is, for example, 1,000, but may be more or less.

Figure 12 is a point _{Sa i (i = 1,2, ...} , N) belonging to the point group GSa of an example of distribution on a two-dimensional plane. Point Sa _i in the illustrated example are distributed annularly.
As described above, a number of coordinates Sa i _(p, q) determined for the longitudinal direction position point group GSa consisting are stored in the coordinate storage unit 46.
The stored point cloud GSa may be continued to be used as long as the inspection target is a steel rope having the same appearance or surface shape, for example, a steel rope having the same specifications.

In the inspection execution mode, after acquiring surface data, calculating similarity, and calculating coordinates for the position of interest of the steel cable 11 to be inspected as described above, the stored point cloud GSa is used. , Calculate the isolation index.

That is, the isolated index calculation unit 47 calculates the point _{S k} for the target position, the point _{Sa i (i = 1,2, ...} , N) isolated index is divergence degree for point cloud GSa consisting.
Figure 12 also shows an example of a point S _k for the target position. In the illustrated example, the point S _k is slightly away from the circular ring formed by the point Sa _i.

Solitary index, among the points Sa _i of the point group GSa, relatively close to a point _{S k,} to each of the points Sa _i number (Nc) of predetermined, based on the sum of the distance from a point _{S k} Is calculated.
The isolation index is calculated by, for example, the following method.

Point _{S k} and the point _{Sa i (i = 1,2, ...} , N) the distance on a two-dimensional plane and each of _{d i (i = 1,2, ...} , N) is calculated.

Then, the distance _{d i (i = 1,2, ...} , N) Nc pieces from smaller of (Nc is the predetermined rata number, Nc«N: for example Nc = 10 when N = 1000) To extract.
For example, the distance _{d i (i = 1,2, ...} , N) distance to sort in ascending order _{e i (i = 1,2, ...} , N) was obtained, and the distance _e i (i = 1,2 , ..., N) are extracted from the smaller ones.

Next, the total AL (k) of the extracted distance e _i (i = 1, 2, ..., Nc) is calculated. The sum AL (k) is intended to represent the degrees of deviation for the point group GSa of the point S _k, it is used as a solitary index.

The above processing is represented by the following formulas (3a) and (3b).

Abnormality determination unit 48 determines whether there is an abnormality in the linear portion of the longitudinal direction position corresponding to each point S _k based on solitary index calculated by the solitary index calculation unit 47 AL (k), determination The result AB (k) is output.
For example, if the isolation index AL (k) is larger than a predetermined threshold value, it is determined that there is an abnormality in the linear portion at the position in the long direction from which the surface data is obtained.

It should be noted that it may be determined whether or not there is an abnormality based on the isolation index calculated for the linear portions of two or more continuous long direction positions.
For example, it may be determined that there is an abnormality when the isolation index for the linear portion of a predetermined number of continuous long direction positions is larger than the threshold value.

By calculating the isolation index as described above and determining the presence or absence of an abnormality based on the calculated isolation index, it is possible to reduce the error in determining the presence or absence of an abnormality.

For example, as described with reference to FIG. 11, even when a perfect circular locus cannot be obtained, it is possible to appropriately determine the presence or absence of an abnormality.

Although the abnormality detection unit 35-1 has been described above, the same processing is performed in the abnormality detection units 35-2 to 35-4.
The inspection result output unit 36 receives the determination result AB (k) output from the abnormality detection unit 35-1 to 5-4, aggregates it, and outputs it to the outside as an inspection result. For example, the inspection result may be displayed by outputting to a monitor (not shown). Alternatively or additionally, it may be output to a printer (not shown) to print the inspection result.
Instead of providing the inspection result output unit 36, a storage unit for storing the inspection result may be provided so that the inspection result can be read out later when needed. The storage unit may be configured with a removable memory. The inspection result output unit 36 and the above-mentioned storage unit may be provided together.

The isolation index calculation unit 47 calculates the isolation index based on the distance between the point about the position of interest and the point belonging to the point cloud.
Instead of the above distance, the isolation index may be calculated using a function that monotonically increases or decreases with respect to the distance.

When a function that increases monotonically is used, when selecting the above-mentioned predetermined number (Nc) points, it may be selected from the smaller ones as described above. When using the monotonically decreasing function, when selecting the above-mentioned predetermined number (Nc) points, it is necessary to select from the larger ones, contrary to the above.
Furthermore, the solitary index calculation unit 47, when calculating the solitary index with monotonically decreasing function with respect to the distance, solitary index to be calculated, the S _k points obtained for the target position, The larger the degree of deviation from the point cloud, the smaller the value. Therefore, the abnormality determination unit 48 determines that there is an abnormality when the isolation index is smaller than a predetermined threshold value.

The reference data storage unit 42 stores the sinusoidal reference data Ra and Rb.
The reference data Ra and Rb may have waveforms of other shapes. For example, it may have a waveform that changes in the same manner as a change in the pixel value constituting the surface data Dk.

According to the first embodiment, _{the isolation index is calculated based on the distance between the point Sk} obtained for the attention position and the points constituting the point cloud, and the presence or absence of an abnormality is determined based on the calculated isolation index. Therefore, it is possible to reduce an error in determining the presence or absence of an abnormality.

Embodiment 2.
The abnormality detection unit used in the inspection device 30 of the second embodiment is the same as the abnormality detection unit 35-1 shown in FIG.

In the first embodiment, the point group generation mode, the point _{Sa i (i = 1, ...} N) to form a point cloud GSa by accumulating, in the inspection execution mode, using the point cloud GSa formed, The isolation index is calculated.
_{In the second embodiment, the points S i} obtained after the start of the process in the inspection execution mode are accumulated, _{and the isolated point S i} is calculated using the accumulated points S i. In other words, the point cloud is generated in parallel with the inspection. In this case, among the accumulated points S _i, constitute a point group in terms other than the point corresponding to the surface data of the target position, for point cloud constructed, deviance points corresponding to the surface data of the target position Just ask.

For example, it is assumed that the surface data is obtained by sequentially taking images at different positions in the long direction, and the number of points constituting the point cloud is N. In that case, a point cloud is formed at the points of the coordinates calculated from the surface data of the N long-length directional positions acquired by N times of imaging, which are performed prior to the imaging at the attention position, and attention is paid. The degree of deviation of the coordinate points calculated from the surface data of the position with respect to the above point cloud is calculated as an isolation index.

In this case, each time the attention position changes, the oldest of the above N long-distance positions is excluded, and instead, the position that was the attention position one time before is one of the N long-distance positions. It will be included in one.

When the processing is performed as described above, the isolation index may be calculated by excluding the points obtained from the surface data for the position in the long direction close to the position of interest. For that purpose, for example, among the above N long-length directional positions, the points obtained from the surface data of the F long-long directional positions (F is a natural number of 1 or more) that were recently imaged are obtained. You can exclude it.

Similar to the process of the first embodiment, the above process is performed after acquiring surface data for a large number of linear portions in the longitudinal direction, calculating the first and second similarities, and calculating the coordinates. It can be said that this is a process of acquiring surface data for the linear portion of the attention position, calculating the first and second similarities, calculating the coordinates, and calculating the isolation index.

The same effect as that of the first embodiment can be obtained in the second embodiment. Further, since the coordinate group (point cloud) is generated after the start of the inspection, the inspection can be performed even if the coordinate group (point cloud) is not generated in advance.

Embodiment 3.
The abnormality detection unit used in the inspection device 30 of the third embodiment is the same as the abnormality detection unit 35-1 shown in FIG.

In the second embodiment, after the processing in the inspection execution mode is started, surface data is acquired for a large number of linear portions in the long direction, the first and second similarities are calculated, and the coordinates are calculated. Later, the surface data for the linear portion of the attention position is acquired, the first and second similarities are calculated, the coordinates are calculated, and the isolation index is calculated.
In the third embodiment, after the processing in the inspection execution mode is started, surface data is acquired for linear portions at different positions in the long direction, first and second similarities are calculated, and coordinates are calculated. After that, one of the above-mentioned different positions is selected as the attention position, and the positions other than the selected positions are set as the above-mentioned many positions, and the isolation index is calculated.

For example, the points of the coordinates of the first number calculated based on the surface data of the linear portion of the linear portion of the long direction position of the predetermined number are accumulated, and the accumulated points of the first number are accumulated. select S _i in order to constitute a point group in (S _k other than those of the point S _i) second number of points of other than S _k selected points, the S _k points selected for the point group The degree of divergence may be calculated as an isolation index. The above-mentioned first number is a number obtained by adding 1 to the above-mentioned second number.
The first number above is, for example, the number of longitudinal positions from which surface data can be obtained while the steel rope travels by a predetermined length (eg, 10 m).
Such processing can be called bulk processing.

Hereinafter, represent _{S i} point accumulated in bulk processing by symbol Sb _i, representing the composed point group by symbol GSb at point Sb _i.
Even when bulk processing is performed, the isolation index is calculated by excluding the points obtained from the surface data for the long direction position close to the long direction position (attention position) corresponding to the selected point. Is desirable.

Hereinafter, the procedure for calculating the isolation index in the case of performing the above bulk processing will be described with reference to FIG. 13.
Processing shown in FIG. 13, N (N is the predetermined number) point Sb i of _(i = 1, ... N) is started each time obtained. The numbers from 1 to N are assigned to each linear portion according to the position in the long direction of the steel rope. In the following, when surface data is obtained from one side to the other in order for a predetermined length (length in the long direction) portion of the steel rope, the numbers from 1 to N represent the order in which the surface data is obtained.

The isolation index is calculated for the attention position in the surface data whose numbers are 1 to N, with the long direction positions other than the fixed number of long direction positions near both ends as the attention positions in order.

In step ST11, k is set to the initial value of 1 + F. F is a predetermined number.
In step ST12, the point _{Sb i (i = 1,2, ...} , N) from among, for selecting a point _{S k.}

In step ST13, the point _{Sb i (i = 1,2, ...} , N) among points of _{Sb i (i = k-F} , k-F + 1, k-F + 2, ..., k + F-2, k + F-1, k + F) Extract points other than.
The extracted points are represented by Sc _i (i = 1, 2, ..., N-2F-1).
In step ST14, the point _{S k} and the point _{Sc i (i = 1,2, ...} , N-2F-1) distance on a two-dimensional plane and each of _{d i (i = 1,2, ...} , N- 2F-1) is calculated.

In step ST15, the distance _{d i (i = 1,2, ...} , N-2F-1) rearranging the smaller ones in order. The distance after sorting _{is represented by e i} (i = 1, 2, ..., N-2F-1).
In step ST16, _{Nc (Nc << N-2F-1) are extracted from the smaller distance e i} (i = 1, 2, ..., N-2F-1), and the sum of the extracted distances is obtained. ..

Combinations of steps ST15 and ST16, the distance _{d i (i = 1,2, ...} , N-2F-1) extracting the Nc number (Nc«N-2F-1) from the small of the extracted Nc number It can be said that it is a process of obtaining the total AL (k) of the distances of.

Total AL calculated in step ST16 (k) is used as solitary indicator of the divergence degree for the point group GSb the point _{S k.}
The processing of steps ST15 and ST16 is represented by the following equations (4a) and (4b).

In step ST17, it is determined whether or not k is equal to n + F.
If they are not equal, the process proceeds to step ST18. In step ST18, 1 is added to k. After step ST18, the process returns to step ST12.
If k is equal to n + F in step ST17, the process ends.

The above processing excludes the F linear portion near one end and the F linear portion near the other end from the N linear portions (ST11, ST17), and after exclusion. of (N-2F) selecting a linear portion of the pieces in the order (ST12), a point for the selected linear portion as S _k (ST12), around the selected linear portions, F th previous Exclude from the linear part to the linear part after F (ST13), and calculate the total distance between the points for the excluded linear part and the points for the selected linear part (ST14, ST15, ST16), it can be said that it is a process of outputting the sum of distances as an isolation index.

14, the point Sb _i when performing the bulk process, an example of distribution on a two-dimensional plane. 14, of the point Sb _i indicates the point _{S k} by "*" indicates the point _{Sb k-F ~Sb} k ₊ F a ^"." _Indicates the _{Sc 1} ~Sc _N-2F-1 in the "-" ..

Processing in step ST13 is less than a certain solitary index points S _k of coordinates calculated for the linear portion, the influence of the point of coordinates calculated for the other linear portion located in the vicinity of the linear portion This is to prevent it from being calculated as a value.
For example, if the interval between the linear portions of the surface data is acquired by the imaging is narrow, the point S _i is plotted at a high density on a two-dimensional plane. If the range of the abnormality extends to two or more linear parts and there is an abnormality over a certain linear part and the linear part in the vicinity thereof, the point corresponding to the surface data of the certain linear part and the position in the vicinity thereof. The distance from the point corresponding to the surface data of the other linear portion becomes small, and the isolation index becomes a small value. By excluding the points about the surface data of the linear part located near the above-mentioned linear part and calculating the isolation index based on the distance from the points about the linear part other than the neighboring linear part. , It is possible to reduce an error in determining the presence or absence of an abnormality in the point corresponding to the surface data of the linear portion.

In the procedure shown in FIG. 13, of the N linear portions, the F linear portion near one end and the F linear portion near the other end are excluded (ST11, ST17), such exclusion is not always necessary. That is, all of the N linear portions may be selected in order.
This is because in step ST13, the selected linear portion is centered, and the linear portion before F pieces to the linear portion after F pieces are excluded.

As described above, the same effect as that of the first embodiment can be obtained in the third embodiment. Further, since the isolation index is calculated by bulk processing, the inspection can be performed even if the coordinate group (point cloud) is not generated in advance.

Embodiment 4.
In the first to third embodiments, the data prepared in advance are used as the reference data Ra and Rb.
In the fourth embodiment, the two surface data obtained by imaging the steel cord are referred to as reference data as in the case of the acquisition of the surface data for generating the point cloud and the acquisition of the surface data Dk for the position of interest. Used as Ra and Rb.

Specifically, the surface data obtained by imaging the linear portions of the steel rope or another steel rope 11 to be inspected at two different positions in the long direction are used as reference data Ra and Rb. ..
It is desirable that the linear portions at the above two positions are not abnormal.
For example, it may be confirmed by directly visually checking the steel rope that there is no abnormality, or the image obtained by imaging the steel rope may be displayed on a monitor (not shown) and visually confirmed by the displayed image. good. In addition to the above methods, if it is guaranteed that there is no abnormality by some means, it may be determined that there is no abnormality based on the guarantee.

For example, the surface data obtained from the linear portion of a certain position (first position) in the elongated direction is used as the first reference data Ra, and the linear portion of another position (second position) in the elongated direction is used. The surface data obtained from the portion and having an orthogonal relationship with the first reference data Ra is used as the second reference data Rb.
In this case, whether or not there is an orthogonal relationship is determined by whether or not the relationship of the above equation (1) is satisfied.

FIG. 15 shows an abnormality detection unit 35-1b used in the inspection device 30 of the fourth embodiment.
The illustrated abnormality detection unit 35-1b includes a data acquisition unit 41b instead of the data acquisition unit 41 in FIG.

The inspection device 30 of the fourth embodiment operates in the reference data generation mode prior to the processing in the inspection execution mode.

In the reference data generation mode, the data acquisition unit 41b is a linear portion of the steel rope or another steel rope 11 to be inspected at two different positions in the longitudinal direction, that is, the first position and the second position. The surface data obtained by imaging the image is output as reference data Ra and Rb.
If the camera 25-1 is an area camera, surface data generated from two different rows of imaged data in the captured image may be used as reference data Ra and Rb.
If the camera 25-1 is a line camera, surface data generated from two imaging data obtained by imaging at two different timings may be used as reference data Ra and Rb. For example, the surface data generated from the imaging data acquired at a certain timing is used as the first reference data Ra, and then the surface data is generated from the imaging data acquired at the timing when the steel cable 11 moves by a predetermined length. The surface data may be used as the second reference data Rb.

FIG. 16 shows the reference data Ra and Rb generated by the above method. In FIG. 16, Tk represents a period.
The example of FIG. 16 assumes a case where the surface data obtained by removing the bias from the imaging data is used as the reference data Ra and Rb.
The reference data Ra and Rb output from the data acquisition unit 41b are stored in the reference data storage unit 42.

The stored reference data Ra and Rb are used to generate a point cloud and perform an inspection.
The point cloud may be generated before the inspection is performed as described in the first embodiment, or may be generated after the start of the processing in the inspection execution mode as described in the second and third embodiments. good. In the latter case, bulk processing may be performed as described in the third embodiment.

When the point cloud is generated before the inspection is performed, the inspection device 30 operates in the point cloud generation mode prior to the processing in the inspection execution mode.

In the point cloud generation mode, as in the first embodiment, the peripheral surface of the steel rope 11 is imaged while moving the steel rope or another steel rope 11 to be inspected, so that the surface is sequentially surfaced for a large number of long direction positions. to obtain data Dk, the reference data Ra stored in the reference data storage unit 42 and the surface data Dk, the similarity p and Rb, the q sought, similarity p, coordinates from q _Sa i (p, q) Ask for.
In this case, it is desirable that the portion of the steel cable 11 to be imaged is a portion having no abnormality.
Acquisition of surface data Dk, the calculation and the coordinate calculation of similarity, different number of (N number of) multiple points Sa i by performing the longitudinal direction position of steel ropes 11 _(i = 1,2, ..., A point cloud GSa consisting of N) is formed.

FIG. 17 shows an example of the point cloud GSa obtained by the above processing.
The locus drawn by the point cloud GSa shown in FIG. 17 has a larger degree of deviation from the perfect circle than the locus shown in FIGS. 10, 11 and 12.
The point cloud GSa is stored in the coordinate storage unit 46.

In the inspection execution mode, as in the first embodiment, the peripheral surface of the steel rope 11 is imaged while moving the steel rope 11 to be inspected, so that the linear shape of each of the different positions of the steel rope 11 in the long direction is obtained. The surface data Dk is acquired for the portion, the similarity p and q between the acquired surface data Dk and the reference data Ra and Rb stored in the reference data storage unit 42 are calculated, and the similarity p and q are two-dimensional. Find the coordinates _Sk (p, q).

The calculation of the isolation index and the determination of the presence or absence of an abnormality can be performed in the same manner as described in the first embodiment.
The operation when the point cloud is generated or updated by the bulk processing is the same as that described in the third embodiment.

According to the fourth embodiment, it is possible to correctly determine the presence or absence of an abnormality even when information on how the appearance of the steel rope is changed and the cycle of the change cannot be obtained in advance.

Embodiment 5.
In the fourth embodiment, two surface data obtained by imaging linear portions of the steel rope 11 at two positions in the long direction are used as reference data Ra and Rb.
In the fifth embodiment, a pair of pixel value vectors orthogonal to each other are generated from the surface data obtained by imaging the linear portion of the steel cable 11 at one position in the long direction, and these are referred to as reference data Ra. Used as Rb.

For example, in the description of the fourth embodiment, a pair of pixel value vectors orthogonal to each other are generated from the surface data Dr for either the first position or the second position, and these are used as reference data Ra and Rb. You may.
The process of generating a pair of pixel value vectors orthogonal to each other from one surface data can be performed by, for example, the Hilbert transform.

FIG. 18 shows an abnormality detection unit 35-1c used in the inspection device 30 of the fifth embodiment.
The illustrated abnormality detection unit 35-1c includes a data acquisition unit 41c instead of the data acquisition unit 41b of FIG. 15, and further includes a reference data generation unit 51.

The inspection device 30 of the fifth embodiment operates in the reference data generation mode prior to the processing in the inspection execution mode.
In the reference data generation mode, the data acquisition unit 41c selects the surface data Dr obtained by imaging the linear portion at one position in the long direction of the steel rope or another steel rope 11 to be inspected. And output.
It is desirable that the linear portion at one of the above positions is a portion without abnormality.

The reference data generation unit 51 performs Hilbert transform of the surface data Dr output from the data acquisition unit 41c to generate a pair of pixel value vectors orthogonal to each other, and outputs these as reference data Ra and Rb.

If one of the two pixel value vectors obtained by the Hilbert transform matches the surface data Dr input to the reference data generation unit 51, the other of the two vectors obtained by the Hilbert transform and the reference data generation. The surface data Dr input to the unit 51 may be used as reference data Ra and Rb.

Except for the above, the fifth embodiment is the same as the fourth embodiment.

In the fourth embodiment and the fifth embodiment, the surface data acquired for the first position of the steel rope to be inspected or another steel rope 11 is used as one of the first reference data and the second reference data. It is the same in terms of use.

On the other hand, in the fourth embodiment, the surface data acquired for the second position of the steel rope to be inspected or another steel rope 11 is used as the other of the first reference data and the second reference data. Further, in the fifth embodiment, there is a difference that data orthogonal to one of the first reference data and the second reference data is generated and used as the other of the first reference data and the second reference data.

In the fifth embodiment, the same effect as in the fourth embodiment can be obtained. Further, even when a pair of orthogonal surface data cannot be obtained directly from the captured image, a pair of orthogonal reference data can be generated.

Embodiment 6.
In the fourth embodiment, in the reference data generation mode, two surface data obtained by imaging the steel rope or another steel rope 11 to be inspected are stored as reference data Ra and Rb, and then a point cloud. It is repeatedly used to calculate the similarity of surface data Dk for linear portions at different positions, which is acquired for the generation and inspection of the data.

The surface data that is the target of the calculation of similarity and the calculation of coordinates in the generation of point clouds and the implementation of inspection is hereinafter referred to as "measurement data" for convenience. This is to distinguish it from the surface data acquired in the reference data generation mode and used as the reference data. In the point cloud generation mode, points constituting the point cloud are formed from the measurement data. In the inspection execution mode, the isolation index is calculated for the coordinates calculated from the measurement data.

In the sixth embodiment, the acquisition of the measurement data Dk and the acquisition of the reference data Ra and Rb are performed in parallel, and the reference data Ra and Rb are updated every time the measurement data Dk is newly acquired.

Acquiring the measurement data Dk and acquiring the reference data Ra and Rb in parallel is not limited to the case where these acquisitions are performed exactly at the same time, but also a certain time difference immediately before or after the acquisition of the measurement data Dk. This includes the case where the reference data Ra and Rb are acquired.
There may be a time lag between the acquisition of the first reference data Ra and the acquisition of the second reference data Rb.

For example, when the measurement data Dk is acquired from the linear portion of the attention position, the surface data for the first position different from the attention position is used as the first reference data Ra, and the surface data for the second position is used. Is used as the second reference data Rb.
The first position and the second position are relative to each other so that the surface data obtained from these linear portions are orthogonal to each other when the linear portions of these positions are normal. Relationships are defined. Further, either the first position or the second position is set to a position separated by a certain distance from the position of interest.
In this way, it can be said that the first and second positions are positions related to the position of interest.

FIG. 19 shows an abnormality detection unit 35-1d used in the inspection device 30 of the sixth embodiment.
The illustrated abnormality detection unit 35-1d includes a data acquisition unit 41d instead of the data acquisition unit 41b in FIG. The reference data storage unit 42 in FIG. 15 is omitted.

Hereinafter, a case where an area camera is used as the camera 25-1 will be described.
FIG. 20 is a diagram similar to FIG. 7, and shows an example of an image obtained by imaging with the camera 25-1 superimposed on the steel cable 11 to be imaged. FIG. 20 shows the range of the steel rope 11 to be imaged by the dotted frame 14, and shows the outline of the image obtained by the image inside the frame 14. FIG. 20 further shows three pre-designated rows Lra, Lrb, Lk.

The row Lk is a row in which the linear portion of the attention position is imaged.
Rows Lra and Lrb are rows in which the linear portions of the first and second positions of the steel rope 11 are imaged, respectively. That is, when the linear portion of the attention position is imaged in the row Lk, the linear portion of the first position is imaged in the row Lra, and the linear portion of the second position is imaged in the row Lb.

The data acquisition unit 41d outputs the surface data obtained from the above three rows Lra, Lrb, and Lk in the captured image as the first reference data Ra, the second reference data Rb, and the measurement data Dk, respectively. ..

FIG. 21 shows an example of reference data Ra, Rb and measurement data Dk obtained from rows Lra, Lrb, Lk of FIG. In FIG. 21, the vertical axis indicates the brightness of the image, and the horizontal axis indicates the row direction position in the frame 14 of the dotted line in FIG. _{The x l} and x _{r in} FIG. 21 correspond to _{the left end x l} and the right end x _r of the dotted frame 14 in FIG. 20.
_{The x l} and x _{r in} FIG. 21 correspond to _{the left end x l} and the right end x _r of the dotted frame 14 in FIG. 20.

The similarity calculation unit 43 calculates the first similarity p, which is the similarity between the measurement data Dk and the first reference data Ra. The similarity calculation unit 43 also calculates a second similarity q, which is the similarity between the measurement data Dk and the second reference data Rb.
The coordinate calculation unit 45 calculates the two-dimensional coordinates _Si having the first similarity p and the second similarity q as elements.

The data Ra, Rb, Dk are obtained from the fixed rows Lra, Lrb, Lk in the image, and the phases of the data Ra, Rb and the data Dk (phases in the periodic change of the steel cable in the longitudinal direction). the difference in is constant, the similarity p, without q also changes greatly, therefore, not be greatly changed even coordinates S _i. Therefore, the point _Si is located on a two-dimensional plane within a narrow range, for example, as shown in FIG.

The generation of the point cloud used for the calculation of the isolation index may be performed before the inspection as described in the first embodiment, and the inspection may be performed as described in the second and third embodiments. It may be performed after the processing in the mode is started. In the latter case, bulk processing may be performed as described in the third embodiment.

When the point cloud is generated before the inspection is performed, the inspection device 30 operates in the point cloud generation mode prior to the processing in the inspection execution mode.

In the point cloud generation mode, the peripheral surface of the steel rope 11 is imaged while moving the steel rope or another steel rope 11 to be inspected, so that the measurement data Dk is sequentially acquired at a large number of long direction positions and each of them is obtained. When the measurement data is acquired at the long direction position, the reference data Ra and Rb are acquired at the positions related to the long direction position, and the measurement data Dk and the length acquired for each long direction position are acquired. reference data Ra obtained at the position associated with the longitudinal direction position, obtains the similarity p, q and Rb, obtains the similarity p, coordinates from _{q Sa i (p, q)} .
In this case, it is desirable that the portion of the steel cable 11 to be imaged is a portion having no abnormality.

Acquisition of the measurement data Dk, the reference data Ra, acquisition of Rb, the calculation and the coordinate calculation of similarity, the wire rope 11 different number of mutually (N number of) multiple points Sa _i by performing the longitudinal direction position ( A point cloud GSa consisting of i = 1, 2, ..., N) is formed.

FIG. 23 shows an example of the point cloud GSa obtained by the above processing.
In FIG. 23, shows the point Sa _i in ".".
Point Sa _i as shown, are concentrated in a narrow range on the ring. The range of concentration depends on the positional relationship between the linear portions for acquiring the surface data of the steel rope 11, and therefore the positional relationship between the rows Lra and Lrb and the rows Lk in FIG.

In the inspection execution mode, while moving the steel rope 11 to be inspected, the measurement data Dk is acquired at each of the different positions in the elongated direction of the steel rope 11, and the measurement data is acquired at that position. , The reference data Ra and Rb are acquired at the position related to the position, and the similarity between the measurement data Dk acquired at the position and the reference data Ra and Rb acquired at the position related to the position. P and q are calculated, and the two-dimensional coordinates _Sk (p, q) are obtained from the similarity p and q.
In FIG. 23, the point _Sk is indicated by “*”.

The calculation of the isolation index and the determination of the presence or absence of an abnormality can be performed in the same manner as described in the first embodiment.

The operation when the point cloud is generated or updated by the bulk processing is the same as that described in the third embodiment.
Figure 24 is the point Sb _i accumulated in bulk process, an example of distribution on a two-dimensional plane.
In Figure 24, of the point Sb _i indicates the point _{S k} by "*" indicates the point _{Sb k-F ~Sb} k ₊ F a ^"." _Indicates the _{Sc 1} ~Sc _N-2F-1 in the "-" ..

Point Sb _i as shown in FIG. 24, as with Sa _i point shown in FIG. 23 is concentrated in a narrow range on the ring. The range of concentration depends on the relationship between the positions where the data is acquired, that is, the positional relationship between the rows Lra and Lrb in FIG. 20 and the rows Lk.

The calculation of the isolation index and the determination of the presence or absence of an abnormality can be performed in the same manner as described in the third embodiment.

The case where the camera 25-1 is an area camera and the surface data acquired in the rows Lra and Lrb different from the row Lk in which the measurement data Dk in the image is acquired is used as the reference data Ra and Rb has been described above.

When the camera 25-1 is a line camera, surface data acquired at different timings may be used as data Ra, Rb, and Dk, respectively. For example, the surface data acquired at a certain timing is used as the first reference data Ra, and then the surface data acquired at the timing when the steel cable 11 moves by a predetermined length (first length) is used as the second surface data. The surface data may be used as the reference data Rb of the above, and then the surface data acquired at the timing when the steel cable 11 moves by a predetermined length (second length) may be used as the measurement data Dk.
The first length and the second length described above may be different from each other or may be the same.

In the present embodiment, the calculation of the isolation index is performed by the formulas (3a) and (3b) or the formulas (4a) and (4b) as described in the first or third embodiment. However, it is also possible to calculate the isolation index by other methods.
For example, when the distribution of the point group on the two-dimensional plane is as shown in FIGS. 22, 23, and 24 and can be regarded as a two-dimensional normal distribution, attention is paid by analyzing the distribution characteristic of the point group. calculating a Mahalanobis distance for the point group of the points S _k for the position, and calculates the Mahalanobis distance calculated may be used as an isolated index of the point S _k.

According to the sixth embodiment, as in the fourth embodiment, it is possible to correctly determine the presence or absence of an abnormality even when information on how the appearance of the steel rope is changed and the cycle of the change cannot be obtained in advance.
In addition, since the measurement data and the reference data are acquired in parallel, it is possible to appropriately determine the presence or absence of an abnormality even when the appearance changes depending on the position of the steel rope in the long direction. ..

Embodiment 7.
In the sixth embodiment, two surface data obtained by imaging linear portions of the steel rope 11 at two positions in the long direction are used as reference data Ra and Rb.
In the seventh embodiment, a pair of pixel value vectors orthogonal to each other are generated from the surface data obtained by imaging the linear portion of the steel cable 11 at one position in the long direction, and these are referred to as reference data Ra. Used as Rb.

For example, a pair of pixel value vectors orthogonal to each other are generated from the surface data Dr for either the first position or the second position in the description of the sixth embodiment, and these are used as reference data Ra and Rb. Use.
The process of generating a pair of pixel value vectors orthogonal to each other from one surface data can be performed by, for example, the Hilbert transform.

FIG. 25 shows the abnormality detection unit 35-1e used in the inspection device 30 of the seventh embodiment.
The illustrated abnormality detection unit 35-1e includes a data acquisition unit 41e instead of the data acquisition unit 41d in FIG. 19, and further includes a reference data generation unit 51.

Similar to the sixth embodiment, when the camera 25-1 is an area camera and the image obtained by the image taken by the camera 25-1 is as shown in FIG. 20, one of the rows Lra and Lrb of FIG. 20. For example, the row Lra is designated as the row for acquiring the above surface data Dr, and the row Lk is designated as the row for acquiring the above surface data Dk.

The data acquisition unit 41d outputs the surface data obtained from one of the rows Lra and Lrb in the captured image, for example, the row Lra, as the surface data Dr for generating the reference data, and outputs the surface data obtained from the row Lk. , Output as measurement data Dk.

The reference data generation unit 51 performs Hilbert transform of the surface data Dr output from the data acquisition unit 41e to generate a pair of pixel value vectors orthogonal to each other, and outputs these as reference data Ra and Rb.

If one of the two pixel value vectors obtained by the Hilbert transform matches the surface data Dr input to the reference data generation unit 51, the other of the two vectors obtained by the Hilbert transform and the reference data generation. The surface data Dr input to the unit 51 may be used as reference data Ra and Rb.

Except for the above, the seventh embodiment is the same as the sixth embodiment.

In the seventh embodiment, the same effect as in the fifth embodiment can be obtained. Further, even when a pair of orthogonal surface data cannot be obtained directly from the captured image, a pair of orthogonal reference data can be generated.

According to the first to seventh embodiments, in the inspection of the steel rope, it is possible to efficiently detect the occurrence of local rust, fraying, adhesion of foreign matter, etc. on the steel rope, and an error in the determination regarding the presence or absence of an abnormality due to these can be made. Can be reduced.

Modification example.
Various modifications can be added to the above embodiments 1 to 7.
For example, the modifications described for the first embodiment can also be applied to the second to seventh embodiments.

In the above embodiments 1 to 7, the steel rope moves upward. The steel rope may move downward.

In the above embodiments 1 to 7, the first and second reference data that are orthogonal to each other are used, but there is a perfect orthogonal relationship between the first reference data and the second reference data. Even if there is no such relationship, it suffices if there is a relationship close to an orthogonal relationship.

In the above embodiments 1 to 7, the correlation coefficient ρ _a between the surface data Dk and the first reference data Ra is used as the first similarity p, and the surface data Dk is used as the second similarity q. And the second reference data Rb and the correlation coefficient ρ _b are used. As the first similarity p and the second similarity q, those other than the correlation coefficient can also be used.

For example, the reciprocal of the angle formed by the vector constituting the surface data Dk and the vector constituting the first reference data Ra is used as the first similarity p, and the vector constituting the surface data Dk and the second reference data are used. The reciprocal of the angle formed by the vector constituting Rb may be used as the second degree of similarity q.

Further, the inner product of the vector constituting the surface data Dk and the vector constituting the first reference data Ra is used as the first similarity p, and the vector constituting the surface data Dk and the second reference data Rb are configured. The inner product with the vector to be used may be used as the second similarity q.

Further, the reciprocal of the difference between the vector constituting the surface data Dk and the vector constituting the first reference data Ra is used as the first similarity p, and the vector constituting the surface data Dk and the second reference data Rb are used. The reciprocal of the difference from the vector constituting the above may be used as the second degree of similarity q.

Further, the covariance of the vector constituting the surface data Dk and the vector constituting the first reference data Ra is used as the first similarity p, and the vector constituting the surface data Dk and the second reference data Rb are used. The covariance with the vector constituting the above may be used as the second degree of similarity q.

In the above embodiments 1 to 7, the isolation index is calculated based on the distance on the two-dimensional plane with the first similarity p on the horizontal axis and the second similarity q on the vertical axis. There is.
_{Instead, the deviation angle θ and the driving diameter r when the position of the point S i} on the two-dimensional plane (p−q plane) with the above similarity p and q as the coordinate axes are expressed in polar coordinates are obtained, and the deviation angle is obtained. one axis theta, for example, taken on a horizontal axis, to convert the radius vector r other axis, for example, a two-dimensional plane on the vertical axis (theta-r plane) coordinates of S _i of the above, the θ- An isolation index may be calculated based on the distance between points on the r-plane.

The declination θ and the radius r from the similarity p and q can be calculated by the following equations (5a) and (5b).

上記の式においてａ_１及びａ_２は係数であり、例えば予め定められる。The annular locus shown in FIG. 10 is converted into a straight line extending parallel to the θ axis on the θ−r plane. In this case, the respective S _i points constituting point group, the distance d _i between S _k point for the target position, the difference Δθ in the deflection angle theta, but multiplied by the respective coefficient to the difference Δr of the radius vector r It may be defined as the sum of squares. That distance d _i may be defined by the following equation (6).

In the above formula, a ₁ and a ₂ are coefficients, and are predetermined, for example.

The isolation index AL based on the distance d obtained by the formula (6) can be calculated by the above formulas (3a) and (3b) or the calculations of the formulas (4a) and (4b).

In short, the coordinate calculation unit has the coordinates having the first similarity (p) and the second similarity (q) as elements, or the first similarity (p) and the second similarity (q) as variables. Anything may be used as long as it calculates the coordinates having the two functions (θ, r) as elements.

In the above-described embodiments 1 to 7, each of the cameras 25-1 to 25-4 images the peripheral surface of the steel cable and acquires surface data from the imaged data.
A profile measuring instrument may be used instead of the camera. By using the profile measuring instrument, it is possible to detect the unevenness of the steel rope by the strand, that is, the surface shape, and obtain the surface data showing the detection result.
In short, a sensor that can detect the appearance or the surface shape and output the surface data may be used even if it is not a camera or a profile measuring instrument.

In the first to seventh embodiments, the inspection target is a steel cord. The inspection target is not limited to steel ropes. For example, it may be a transmission line for electric power, a fiber rope, or the like.
Further, it may be tubular, rod-shaped, or strip-shaped. Further, it may be in the form of a sheet. These may have a change in appearance or may have a change in surface shape.
There is a canvas as a sheet-like material whose surface shape changes.
Furthermore, the inspection target may be a part of an object provided with many members, parts, etc., such as an apparatus and a structure. For example, if the device or structure has a portion in which at least one of the appearance and the surface shape changes periodically in at least one direction, the portion may be inspected.

In short, the inspection target may be an object or a part of an object in which at least one of the appearance and the surface shape changes periodically in at least one direction.
The change in appearance referred to here includes a change in surface color. Color changes include at least one change in hue, saturation, and lightness. The change in appearance may also be a change in the presence or absence of gloss, the degree of gloss, the presence or absence of transparency, and the degree of transparency.

Changes in surface shape include changes in dimensions when viewed from the observation direction, and changes in the angle formed by the surface with respect to the observation direction. As the surface shape changes, the appearance when viewed from the observation direction may change. When the change in surface shape appears as a change in appearance, it is possible to detect an abnormality based on the change in appearance by imaging with a camera.
A profile measuring instrument can be used to directly measure changes in surface shape.

Even when the inspection target is other than the steel rope, the formation of the coordinate group may be performed based on the surface data obtained from the inspection target itself, and other products or processed products having the same appearance or surface shape as the inspection target or the same thereof. It may be performed based on the surface data obtained from the portion.
Further, in the configurations described in the fourth and fifth embodiments, the reference data may be generated based on the surface data obtained from the inspection target itself, and has the same appearance or surface shape as the inspection target. It may be performed based on the surface data obtained from the product or the processed product or a part thereof.

In the above embodiments 1 to 7, the cameras 25-1 to 25-4 are fixed, and the steel rope moves to realize the relative movement between the steel rope and the camera.
Alternatively, the steel rope may be fixed and the camera may move.

For example, when the steel rope is the steel rope of a crane, the drone equipped with the camera may be flown so that the drone moves in the long direction of the steel rope while keeping the distance from the camera to the steel rope constant. In this case, imaging may be performed at different positions in the long direction.

The drone may also be equipped with a part other than the camera of the inspection device, that is, a processing circuit.
Wheels may be provided on the drone, and the wheels may be brought into contact with the peripheral surface of the steel rope and rolled so that the distance from the camera to the peripheral surface of the steel rope is kept constant.

The moving mechanism does not have to be a dedicated moving mechanism for inspection. For example, when a mechanism for moving the steel rope is provided for other purposes, it is not necessary to provide a dedicated moving mechanism for inspection. As an example, steel ropes may be used in elevator hoisting machines, cable car hoisting machines, and ropeway hoisting machines. The same applies when inspecting steel ropes manufactured on a production line.

In the above-described embodiments 1 to 7, four cameras are provided so as to surround the steel rope, and four abnormality detection units are provided corresponding to the four cameras. However, the number of cameras is not limited to four, and the number of abnormality detection units is not limited to four.
The processing for the image pickup data of the four cameras may be performed in a time-division manner by one abnormality detection unit.
Four cameras cover the entire circumference of the steel rope, but this point is not essential either.
For example, it may be sufficient to inspect only a part of the entire circumference of the steel rope.
In addition, when the inspection target is a band-shaped object, it may be sufficient to take an image with one camera.

As described above, the inspection target may be something other than steel rope.
In short, the inspection target may be an object whose appearance or surface shape changes periodically in at least one direction.

For example, the captured images obtained from the inspection target may be those shown in FIGS. 26, 30, 32 (a), and 36.
These figures show images obtained by imaging with an area camera, and the symbol MV indicates the relative movement direction of the inspection target with respect to the camera (direction in the image corresponding to the relative movement direction).

The image shown in FIG. 26 has a striped pattern in which bright strips and dark strips alternate. The band-shaped portion extends diagonally with respect to the relative moving direction MV. In FIG. 26, the bright part is shown in white and the dark part is shown in a dot pattern.

FIG. 27 shows an example of surface data Dk obtained from one row Lk of the image of FIG. 26. In FIG. 27, the vertical axis indicates the brightness of the image (the size of the pixel value), and the horizontal axis indicates the row direction position in the frame 14 of FIG. 26. _{The x l} and x _{r in} FIG. 27 correspond to _{the left end x l} and the right end x _r of the frame 14 in FIG. 26.
It can be said that the change in the pixel value (change with respect to the row direction position) constituting the surface data shown in FIG. 27 has a rectangular wave shape or a shape close to a rectangular wave shape.

FIG. 28 shows an example of the reference data Ra and Rb used in the calculation of the similarity with respect to the surface data Dk shown in FIG. 27. In the illustrated example, rectangular wavy data (a vector consisting of a sequence of values that change in a rectangular wavy shape with respect to the row direction position) prepared in advance is used.

FIG. 29 is described in the first embodiment using surface data Dk for a large number of linear portions (the surface data shown in FIG. 27 is an example thereof) and reference data Ra and Rb of FIG. 28. and shows an example of the distribution of points S _i obtained by performing the calculation of the similarly similarity calculation and coordinate. In the example shown in FIG. 29, the points _Si are distributed along the sides of a square composed of straight lines inclined by 45 degrees with respect to the p-axis and the q-axis. The center of the square deviates from the origin of the pq coordinate.

Even if the distribution of points S _i is as shown in FIG. 29, it is possible to determine the presence or absence of an abnormality by calculating the isolation index in the same manner as described above.

The image of FIG. 30 has a zigzag pattern. The zigzag pattern is formed by diagonal lines with respect to the relative moving direction MV of the inspection target with respect to the camera. In FIG. 30, bright areas are shown in white, dark areas are shown in cross-hatched areas, and intermediate brightness areas are shown in dot patterns.
Also in the case of the image of FIG. 30, the reference data Ra and Rb shown in FIG. 28 can be used.

In FIG. 31, the similarity is calculated and the coordinates are calculated in the same manner as described in the first embodiment using the surface data Dk for a large number of linear portions and the reference data Ra and Rb in FIG. 28. points obtained in an example of the distribution of S _i. In the example shown in FIG. 31, the points _Si are distributed along a part of the side of the square in FIG. 29.

Even if the distribution of points S _i is as shown in FIG. 31, it is possible to determine the presence or absence of an abnormality by calculating the isolation index in the same manner as described above.

In the examples described with reference to FIGS. 26 to 29 and the examples described with reference to FIGS. 30 to 31, rectangular wavy reference data is used as reference data, but instead, sinusoidal reference data is used. May be used.

In the image of FIG. 32 (a), in each of the band-shaped portions extending diagonally with respect to the relative moving direction MV of the inspection target with respect to the camera, the brightness gradually increases from one edge to the other. Change. For example, if the outline of the change in brightness is shown for the portion along the line Lk, it becomes a sawtooth shape as shown in FIG. 32 (b). As described above, the change in brightness is continuous, but in FIG. 32 (a), the bright part is shown in white, the dark part is shown by cross-hatching, and the intermediate brightness part is shown by dots. It is shown in a pattern.

FIG. 33 shows an example of surface data Dk obtained from one row Lk of the images of FIGS. 32 (a) and 32 (b). In FIG. 33, the vertical axis indicates the brightness of the image (the size of the pixel value), and the horizontal axis indicates the row direction position in the frame 14 of FIG. 32. _{The x l} and x _{r in} FIG. 33 correspond to _{the left end x l} and the right end x _r of the frame 14 in FIG. 32 (a).
It can be said that the change in the pixel value (change with respect to the row direction position) constituting the surface data shown in FIG. 33 has a sawtooth shape or a shape close to a sawtooth wave.

FIG. 34 shows an example of the reference data Ra and Rb used in the calculation of the similarity with respect to the surface data Dk shown in FIG. 33. In the illustrated example, sawtooth data (a vector consisting of a sequence of values that change in a sawtooth shape with respect to the row direction position) is used.

FIG. 35 is described in the first embodiment using surface data Dk for a large number of linear portions (the surface data shown in FIG. 33 is an example thereof) and reference data Ra and Rb of FIG. 34. and shows an example of the distribution of points S _i obtained by performing the calculation of the similarly similarity calculation and coordinate. In the example shown in FIG. 35, the points _Si are distributed along two curves connected to each other at both ends.

Even if the distribution of points S _i is as shown in FIG. 35, it is possible to determine the presence or absence of an abnormality by calculating the isolation index in the same manner as described above.

The image of FIG. 36 has a zigzag pattern similar to that of FIG. 30, but the runout width in the row direction is smaller than that of the image of FIG. 30.
Regarding the image of FIG. 36, as described in the seventh embodiment, the acquisition of the surface data (measurement data) Dk to be the target of the coordinate calculation and the generation of the reference data Ra and Rb are performed in parallel, and the measurement data Dk is performed. Will be described on the assumption that the reference data Ra and Rb are updated every time the data is newly acquired.

As described in the seventh embodiment, the reference data Ra and Rb are generated by Hilbert transforming the surface data Dr acquired from the row Lra or Lrb (for example, Lra) different from the row Lk from which the measurement data Dk is acquired. It is composed of a pair of pixel value vectors orthogonal to each other.

If one of the two pixel value vectors obtained by the Hilbert transform matches the original surface data Dr, the other of the two vectors obtained by the Hilbert transform and the surface data Dr are referred to as reference data Ra. , Rb may be used.

37 (a) shows an example of the measurement data Dk obtained from the row Lk of the image of FIG. 36. FIG. 37B shows an example of reference data Ra and Rb.

FIG. 38 shows measurement data Dk for a large number of linear portions (the data shown in FIG. 37 (a) is an example thereof), and reference data Ra and Rb generated each time the measurement data Dk is acquired. Figure 37 the data shown in (b) are examples) and likewise similarity calculation and distribution of points S _i obtained by performing the calculation of the coordinates of the as described in embodiment 1 with reference to An example is shown. In the example shown in FIG. 38, the points _Si are distributed along the arc.

Even if the distribution of points S _i is as shown in FIG. 38, it is possible to determine the presence or absence of an abnormality by calculating the isolation index in the same manner as described above.

In the above embodiments 1 to 7, the cameras are arranged so that the direction in which the image pickup elements of the line camera are lined up or the direction of the row of the area camera corresponds to the direction perpendicular to the moving direction of the steel cable, and the camera is arranged so as to be viewed from the camera. The imaging data of the linear portion extending in the direction perpendicular to the moving direction of the steel rope is used as the surface data.

In the case of an inspection target other than the steel rope, the relative moving direction of the inspection target and the direction of the linear portion for which the surface data is acquired may be perpendicular to the direction seen from the camera in the same manner as described above.

However, this point is not essential, and the relative moving direction of the inspection target and the linear portion for acquiring surface data may be oblique to the direction seen from the camera.

For example, as shown in FIG. 39, bright band-shaped parts (indicated by white) and dark band-shaped parts (indicated by dot patterns) extending in a direction perpendicular to the relative moving direction MV alternate from the inspection target. When the image of the striped pattern appearing in is obtained, it is effective to make the linear portion Lk from which the surface data is obtained slanted with respect to the relative moving direction MV.
This is because the surface data changes continuously with movement.

In order to make the linear portion Lk for which surface data is obtained slanted with respect to the relative moving direction MV, for example, the direction in which the image sensors of the line camera are lined up or the direction of the row of the area camera is changed to the moving direction MV. It should be slanted with respect to.
Further, in the case of an area camera, the direction of the row may be slanted.
Further, even if the row direction or the column direction is not slanted, the image pickup data consisting of the pixel values obtained from the image pickup elements arranged in the diagonal direction may be acquired as the surface data.

The same can be said when using a profile measuring instrument.

In short, the sensor detects the appearance or surface shape of the linear portion extending in the second direction (different from the first direction) of the inspection target at different positions in the first direction of the inspection target. The data acquisition unit may acquire surface data consisting of a sequence of detection values representing the appearance or surface shape of each linear portion at different positions in the first direction from the detection result of the sensor. All you need is.
Here, the second direction is not always the direction along the straight line. For example, in the case of the steel rope in the embodiment, the second direction is the direction along the peripheral surface.

Furthermore, at least one of the formation of the coordinate group and the generation of the reference data may be performed based on the surface data obtained from another product or processed product having the same appearance or surface shape as the inspection target or a part thereof. I said that. The combination of this and the relationship between the first direction and the second direction is as follows.

The coordinate group used to calculate the isolation index includes detection by the sensor of the linear part of the attention position to be inspected, acquisition of surface data, calculation of the first similarity and the second similarity, and calculation of coordinates. If it is obtained by performing the same treatment on a product or a processed product having the same appearance or surface shape as the inspection target or the inspection target, or linear portions at a large number of positions in the first direction of the portion. good.
With respect to a product or work piece or part thereof having the same appearance or surface shape as the inspection target, the first direction thereof means the direction corresponding to the first direction of the inspection target.

Further, in the configurations described in the fourth and fifth embodiments, as the first reference data, the inspection target or the product or processed product having the same appearance or surface shape as the inspection target or a portion thereof in the first direction. The surface data acquired for the linear portion at the first position may be used.

Further, as the second reference data, a second position different from the first position in the first direction of the inspection target or the product or processed product having the same appearance or surface shape as the inspection target or a part thereof. The surface data acquired for the linear portion of the above may be used.

Although the inspection device of the present invention has been described above, the inspection method carried out by the inspection device also forms a part of the present invention. Further, a program for causing a computer to perform processing in the processing circuit of the inspection device, a recording medium readable by the computer in which such a program is recorded, for example, a non-temporary recording medium also forms a part of the present invention.

11 Steel rope, 12 strands, 20 moving mechanism, 21 frame, 22 rollers, 25-1 to 25-4 cameras, 30 inspection equipment, 32 processing circuits, 35-1 to 35.4, 35-1b, 35-1c, 35 -1d, 35-1e anomaly detection unit, 36 inspection result output unit, 41, 41b, 41c, 41d, 41e data acquisition unit, 42 reference data storage unit, 43 similarity calculation unit, 46 coordinate calculation unit, 46 coordinate storage unit. , 47 Isolation index calculation unit, 48 Anomaly determination unit, 51 Reference data generation unit.

The inspection device of the present invention is
A sensor that detects the appearance or surface shape of a linear portion extending in a second direction different from the first direction at each of different positions in the first direction to be inspected.
One of the different positions is selected as the position of interest, and surface data consisting of a sequence of detection values representing the appearance or surface shape of the linear portion of the position of interest is acquired from the detection result of the sensor. The first similarity, which is the similarity between the surface data and the first reference data, and the second similarity, which is the similarity between the surface data and the second reference data, are calculated, and the first similarity is calculated. And the coordinates having the second similarity as an element, or the coordinates having two functions having the first similarity and the second similarity as variables are calculated.
The same processing as the acquisition of the surface data for the linear portion of the attention position, the calculation of the first similarity and the second similarity, and the calculation of the coordinates is performed on the inspection target or the same appearance as the inspection target. Alternatively, it was obtained by performing on a linear portion extending in the second direction at a plurality of positions in the first direction of a product having a surface shape or a part of the product or a processed product or a part of the processed product. The degree of deviation of the coordinates calculated for the linear portion of the attention position with respect to the group of coordinates is calculated as an isolation index.
It has a processing circuit for determining whether or not there is an abnormality in the inspection target based on the isolation index.
When calculating the isolation index, the processing circuit excludes the coordinates of the linear portion at a position relatively close to the attention position from the coordinates constituting the group of coordinates, and at least the coordinates of the group of coordinates. The isolation index is calculated based on the distance between the coordinates constituting a part of the coordinates and the coordinates of the linear portion of the attention position, or a value that monotonically increases or decreases with respect to the distance.

The inspection method of the present invention is
At each of the different positions in the first direction of the inspection target, the appearance or surface shape of the linear portion extending in the second direction different from the first direction is detected.
One of the different positions is selected as the position of interest, and from the result of the detection, surface data consisting of a sequence of detection values representing the appearance or surface shape of the linear portion of the position of interest is acquired, and the surface data is obtained. The first similarity, which is the similarity between the data and the first reference data, and the second similarity, which is the similarity between the surface data and the second reference data, are calculated, and the first similarity and the first similarity are calculated. Calculate the coordinates having the second similarity as an element, or the coordinates having two functions having the first similarity and the second similarity as variables as elements.
The same processing as the acquisition of the surface data for the linear portion of the attention position, the calculation of the first similarity and the second similarity, and the calculation of the coordinates is performed on the inspection target or the same appearance as the inspection target. Alternatively, it was obtained by performing on a linear portion extending in the second direction at a plurality of positions in the first direction of a product having a surface shape or a part of the product or a processed product or a part of the processed product. The degree of deviation of the coordinates calculated for the linear portion of the attention position with respect to the group of coordinates is calculated as an isolation index.
Based on the isolation index, it is determined whether or not the inspection target has an abnormality.
When calculating the isolation index, the coordinates for the linear portion at a position relatively close to the attention position are excluded from the coordinates constituting the group of coordinates, and at least a part of the group of coordinates is formed. The isolation index is calculated based on the distance between the coordinates and the coordinates of the linear portion of the attention position, or a value that monotonically increases or decreases with respect to the distance.

Claims (17)

A sensor that detects the appearance or surface shape of a linear portion extending in a second direction different from the first direction at each of different positions in the first direction to be inspected.
One of the different positions is selected as the position of interest, and surface data consisting of a sequence of detection values representing the appearance or surface shape of the linear portion of the position of interest is acquired from the detection result of the sensor. The first similarity, which is the similarity between the surface data and the first reference data, and the second similarity, which is the similarity between the surface data and the second reference data, are calculated, and the first similarity is calculated. And the coordinates having the second similarity as an element, or the coordinates having two functions having the first similarity and the second similarity as variables are calculated.
The same processing as the acquisition of the surface data for the linear portion of the attention position, the calculation of the first similarity and the second similarity, and the calculation of the coordinates is performed on the inspection target or the same appearance as the inspection target. Alternatively, the attention to the group of coordinates obtained by performing on the linear portion extending in the second direction at a plurality of positions in the first direction of the product or work piece having a surface shape or the portion thereof. The degree of deviation of the coordinates calculated for the linear part of the position is calculated as an isolation index.
An inspection device having a processing circuit for determining whether or not there is an abnormality in the inspection target based on the isolation index. Each of the first reference data and the second reference data consists of a sequence of values as many as the number of detected values constituting the surface data.
The inspection device according to claim 1, wherein the first reference data and the second reference data have an orthogonal relationship or a relationship close to an orthogonal relationship. The surface data obtained when there is no abnormality in the inspection target is composed of a sequence of detection values that change periodically.
The inspection device according to claim 2, wherein each of the first reference data and the second reference data comprises a sequence of values representing a sine wave having a period equal to or similar to the period of the detected value. The processing circuit acquires surface data about the linear portion of the inspection target or the product or work piece having the same appearance or surface shape as the inspection target or the portion thereof at the first position in the first direction. The inspection device according to claim 2, wherein the acquired surface data is used as one of the first reference data and the second reference data. The processing circuit is linear at a second position different from the first position in the first direction of the inspection target or a product or work piece having the same appearance or surface shape as the inspection target or a portion thereof. The inspection device according to claim 4, wherein surface data for a portion is acquired and the acquired surface data is used as the other of the first reference data and the second reference data. Each time the processing circuit acquires surface data for the linear portion of the attention position, the linear portion of the inspection target at the first position separated from the attention position by a certain distance in the first direction. The inspection device according to claim 2, wherein the surface data of the above-mentioned device is acquired and the acquired surface data is used as one of the first reference data and the second reference data. Each time the processing circuit acquires surface data for the linear portion of the attention position, the processing circuit relates to the linear portion of the inspection target at a second position different from the first position in the first direction. The inspection apparatus according to claim 6, wherein the surface data of the above is acquired and the acquired surface data is used as the other of the first reference data and the second reference data. The processing circuit generates data orthogonal to one of the first reference data and the second reference data and uses it as the other of the first reference data and the second reference data. The inspection device according to 6. After the sensor detects the linear portions at the plurality of positions, the surface data is acquired by the processing circuit, the first and second similarities are calculated, and the coordinates are calculated.
A claim in which the sensor detects the attention position, the surface data is acquired by the processing circuit, the first and second similarities are calculated, the coordinates are calculated, and the isolation index is calculated. The inspection device according to any one of 1 to 8. The sensor is a linear portion extending in a second direction different from the first direction of the inspection target at a predetermined number of positions different from each other in the first direction of the inspection target. Detects appearance or surface shape,
The processing circuit sequentially selects at least a part of the different positions to be the attention position, and forms a group of the coordinates by using a position other than the selected position among the different positions as the plurality of positions. The inspection device according to claim 1. The processing circuit is based on the distance between the coordinates constituting at least a part of the group of coordinates and the coordinates of the linear portion of the attention position, or a value that monotonically increases or decreases with respect to the distance. The inspection device according to any one of claims 1 to 10 for calculating an isolation index. The processing circuit extracts a certain number of distances or values from a small number of values that monotonically increase with respect to the distance or the distance, or is constant from a large value that monotonically decreases with respect to the distance. The inspection device according to claim 11, wherein the values of the number of pieces are extracted and the total of the extracted distances or values is calculated as the isolation index. The processing circuit according to claim 11 or 12, wherein the processing circuit calculates the isolation index by excluding the coordinates of the linear portion at a position relatively close to the attention position from the coordinates constituting the group of coordinates. Inspection equipment. At each of the different positions in the first direction of the inspection target, the appearance or surface shape of the linear portion extending in the second direction different from the first direction is detected.
One of the different positions is selected as the position of interest, and from the result of the detection, surface data consisting of a sequence of detection values representing the appearance or surface shape of the linear portion of the position of interest is acquired, and the surface data is obtained. The first similarity, which is the similarity between the data and the first reference data, and the second similarity, which is the similarity between the surface data and the second reference data, are calculated, and the first similarity and the first similarity are calculated. Calculate the coordinates having the second similarity as an element, or the coordinates having two functions having the first similarity and the second similarity as variables as elements.
The same processing as the acquisition of the surface data for the linear portion of the attention position, the calculation of the first similarity and the second similarity, and the calculation of the coordinates is performed on the inspection target or the same appearance as the inspection target. Alternatively, the attention to the group of coordinates obtained by performing on the linear portion extending in the second direction at a plurality of positions in the first direction of the product or work piece having a surface shape or the portion thereof. The degree of deviation of the coordinates calculated for the linear part of the position is calculated as an isolation index.
An inspection method for determining whether or not there is an abnormality in the inspection target based on the isolation index. Using the detection results obtained by detecting the appearance or surface shape of the linear portion extending in the second direction different from the first direction at each of the different positions in the first direction of the inspection target. An inspection method for inspecting the inspection target.
One of the different positions is selected as the position of interest, and from the detection result, surface data consisting of a sequence of detection values representing the appearance or surface shape of the linear portion of the position of interest is acquired, and the surface data is combined with the surface data. The first similarity, which is the similarity with the first reference data, and the second similarity, which is the similarity between the surface data and the second reference data, are calculated, and the first similarity and the first similarity are calculated. Calculate the coordinates with the similarity of 2 as an element, or the coordinates with the two functions having the first similarity and the second similarity as variables.
The same processing as the acquisition of the surface data for the linear portion of the attention position, the calculation of the first similarity and the second similarity, and the calculation of the coordinates is performed on the inspection target or the same appearance as the inspection target. Alternatively, the attention to the group of coordinates obtained by performing on the linear portion extending in the second direction at a plurality of positions in the first direction of the product or work piece having a surface shape or the portion thereof. The degree of deviation of the coordinates calculated for the linear part of the position is calculated as an isolation index.
An inspection method for determining whether or not there is an abnormality in the inspection target based on the isolation index. A program for causing a computer to execute the process according to the inspection method according to claim 15. A computer-readable recording medium on which the program of claim 16 is recorded.

Images