SE462696B - SET AND DEVICE FOR DETERMINING AN APPROXIMATE, THREE-DIMENSIONAL IMAGE OF A PURPOSE - Google Patents

Description

462 696 10 15 20 25 30 35 2 distinguish objects that in some way deviate from a certain one specific norm. These may be green tomatoes that are folds from the red tomatoes, rusty screws among them clean the screws or vegetables of the wrong kind. Especially in the case of potatoes where potato-like stones may occur it is essential to be able to decide in a simple way what is what.

To be able to scan, measure and determine individuals objects in a large variety of objects are often placed these on a surface or conveyor belt for sorting. In order to achieve fast and economical handling it can be appropriate to at the same time as the object is also identified record the location of the individual object on the surface.

There are several ways to sort objects and in essence, one can distinguish three basic ideas; manual, mechanical and non-contact sorting.

The manual sorting simply requires operation that are adjacent to the surface of the objects on for sorting the objects. In addition to the fact that this is demanding and thus costly, it is inaccurate. In addition involves such temporary work that work environment requirements can be difficult to fulfill.

The mechanical sorting can be performed on many way. There are known ways to feel the weight of the object by means of levers, springs and the like. In this case, however, there is some risk of objects of different varieties, but with the same weight race together. In the case of potatoes are used especially sieve sorting. The potatoes are sorted using a sieve device comprising superimposed sieve with sieve stitches, which in Sweden are square. Potatoes- shake forward on the sieve device until they slip properly. nom a sieve with sieve stitches of suitable size. At last the potatoes reach a sieve, whose sieve meshes are so small that the potatoes can not slip through. If, for example, a potato tis can pass a sieve mask with 55 mm side but not one sieve mask with 35 mm side, the potatoes are of the size 35/55.

There is a certain risk that long and narrow potatoes will be mixed. with small round potatoes, since the elongated potato 10 15 20 25 30 35 4-62 696 3 tisen in its largest cross section has the same dimensions as the round potato. E Mechanical sorting is not completely accurate, can give cause damage to the objects and result in a mixed objects of different sizes.

The non-contact methods can be optical, electrical one way utilizes a camcorder with CCD picture tube for imaging each object. After- that no operator should sit in front of a video screen image analysis is used. In this case, for example, color deviation ings are detected. The problem with this way of sorting object is that image analysis is both difficult and costly and gives an uncertain result.

Swedish patent 8403364-6 states a method and a device for non-contact detection of plants. Light which has been reflected by plants and a surface is collected and separated into two luminous fluxes, each with a narrow wavelength area. An electronic circuit processes the light signals which has been detected by the device and provides an output signal indicating the presence / non-presence of a plant on surface. However, this detection is rather primitive and can not directly provide more detailed information about the plant location or nature etc.

The main object of the present invention is to provide a method and device for contactless scanning of an object that solves the above problems lemen.

A first special purpose is to determine the the shape of the target.

A second purpose is to determine the weight of the object.

A third purpose is to determine the volume of the object.

A fourth purpose is to determine the center of the object. point, which at some ideal geometric shapes of the object may coincide with the mass center of the object.

A fifth purpose is to determine the position of the object. ring on a surface, with respect to any particular reference.

A sixth purpose is to identify objects and especially distinguish different objects with similar shape. 462 696 10 15 20 25 30 35 4 A seventh purpose is to simultaneously scan two targets, which are located on the same surface or different surfaces.

The above objects are achieved in a manner and a order, as defined by claims 1 and 5, respectively. net drawing parts. Preferred embodiments appear of the dependent claims.

The invention is described in more detail in the following with reference to the accompanying drawings.

Fig. 1a shows from above an object on a surface and Fig. 1b shows the same object in side view seen from the line A-A and fig lc the same object in side view seen from the line B-B.

Fig. 2 schematically shows in perspective a preferred one embodiment of the device according to the invention.

Fig. 3 shows the beam path of the device in Fig. 2.

Fig. 4 shows an object illuminated with parallel light and a diagram of the sound reflected by the object. set intensity.

Fig. 5 shows in perspective view from above how a targets adjacent to a surface are swept by light from two different nings.

Fig. 6a is a diagram of the analog signal which generated upon detection of the reflected radiation one, and Fig. 6b its corresponding digital pulse train.

Fig. 7 is a block diagram of the signal processing according to the invention.

Next, reference is made to Figs. 1a-1c. Fig. 1a shows a object l located on a surface 2. In this example the object resembles a stone. Fig. 1a shows the object straight from above, while Fig. 1b and Fig. 1c show side views from lines A-A and B-B respectively.

When optically scanning the object, the object must dimensions are determined. The maximum length of the object is specified with the designation 3a, while the largest width of the object as is perpendicular to its greatest length is the sum of those arrows denoted by 3b and 3c.

If you study the object on which it is projected the line denoted by A-A is the figure 10 15 20 25 30 35 462 696 5 shown in Fig. 1b, realizing that the largest the length shown in this figure and denoted by 6 is not is the true length of the object. An analogous reasoning applies for the projection of the object on the line B-B. In the figure also shows the object's center of mass 7, which in ideal case coincides with the geometric center of the object.

Next, reference is made to Fig. 2. In a preferred embodiment, several objects 1 are located on a surface 2, which in this case may be one of two endless conveyor belts, which feeds the objects through an art according to the invention rudder scanning device 30. A light emitting means 8 comprises at least one light source. A rotating polygon mirror ll with light-reflecting facets the transmitted light. The light is directed by special lenses 14 and l5, That of band 2, a surface 16 and the object 1 reflected light is led to bake to a detector 9 which comprises at least one photodetector, the output signal is routed to an electronic circuit and computer unit l7.

Next, reference is made to Fig. 3. The light emitting the means 8 preferably comprises a laser of helium s k fê-linser. eg a photodiode. The one of the detector 9 neon type. Of course, other light sources can also be used. coat, such as light bulbs, flash discharge tubes or fluorescent tubes.

The main thing is just that a substantially parallel light bundle is achieved. Since lasers in general and heli- the umlaser in particular has very little divergence for the light beam, this is suitable for the present task.

The laser light beam is sent from the laser towards the the polygon mirror ll via a mirror 10. This mirror has either a hole drilled through it for emission of the laser beam or is semi-transparent.

The rotating polygon mirror 11 has in the preferred The embodiment has six facet surfaces that can reflect the light.

When rotating the rotating polygon mirror 11, one is passed facet into the light beam so that the light beam gradually 462 696 10 15 20 25 30 35 6 will illuminate the facet surface of the polygon mirror. Since the angle of incidence of the light beam towards the facet is thereby changed gradually the light spreads in different directions in dependence of where on the facet the incident laser light meets Father. That of the rotating polygon mirror 11 the light is sent to two distribution mirrors 12, which are located above the two surfaces 2. The light reflects of each distribution mirror 12 so that a part of the light is sent mainly straight down towards the surface 2 and some are sent to their respective side mirrors 13.

The light reflected by the side mirror 13 is transmitted in the direction of the surface 16, which is substantially angular. facing the surface 2. To ensure parallel lighting bundle is the side lens l5 of the so-called f9 type. Gradually as the rotating polygon mirror ll is thus rotated a narrow beam of light to be swept over the surface 16, and if the object is in the beam path on the surface 2 as well over this.

The light sent straight by the distribution mirror 12 down towards the surface 2 is also corrected by means of the fG lens 14, which is located above the surface 2. The object will swept from this direction as well.

Next, reference is made to Fig. 4. In Fig. 4, an object 1 on a surface 2 which is illuminated by incident light rays 190-194. The figure shows also a curve of the intensity of the reflected light that can be perceived by an observer who is in it perpendicular above the surface 2, i.e. at the source of the light beam larna 190-194. When the flat surface 2 reflects the light l90, the intensity of a reflected light beam becomes 200 maximum and this is best shown on the left in the figure.

Namely, the incident light beam 190 is reflected in in accordance with optical reflection laws mainly perpendicular back and gives rise to a lot high intensity, which is called white level 22. In the drawn embodiment, the device is inclined slightly in relation to to the vertical plane. This avoids everything too 10 15 20 25 30 35 462 696 7 strong direct reflection from the surface 2l White level 22 maintained as long as light is reflected by the surface 2, but then the light beam len 191 hits an edge on an object 1 it is sent reflected the beam 201 in a completely different direction.

When this edge is encountered, a viewer who is down perpendicularly above the surface 2 to largely not perceive any light, the intensity decreasing ker to a so-called black level 2l. As the incident the light beam 192 is passed over the object 1, the more electric smiles less light to be reflected in such a direction that some of the light can be perceived by a viewer at an just above the surface 2. The light beam 192 hitting the object the goal is mainly perpendicular will give rise to a strong, reflected light beam 202, which of course also depends on the object's reflectivity properties. One white egg could thus give rise to strong, reflected intensity, while a earthy potato only gives a small increase in intensity compared to black level 21. The intensity curve produced at ref- lesson of the object 1 is shown in the figure and is designated 23.

The important lesson illustrated in Fig. 4 is that right at the outer edges of the object it is perceived intensity is minimal and contrasts sharply with white level. This "edge effect" can be accentuated if the surface 2 is made of a material with strong reflectivity, such as a metal strip or a white plastic mat.

Fig. 4 also shows with a dashed line a reference purification level 40 below which a safe minimum level, ie a river black level, can be determined.

Next, reference is again made to Fig. 3. That part of it reflected the light which as above can be perceived by the viewer perpendicular above the surface 2 is captured with partly the lenses 14, 15 and are reflected by the the mirror 12 to the rotating polygon mirror 11 and on to the mirror l0 where the reflected light be angled towards a detector 9.

In the preferred embodiment, one is sent 462 696 10 l5 20 25 30 35 8 narrow light beam out from the light source 8 and swept with by means of the rotating polygon mirror ll over the object shown on the right side of the figure using halftone beam path. The reflected light will go almost exactly the same way back, i.e. V the same way as the incident light except at the mirror 10 when deflected towards the detector 9.

In the embodiment shown with six facets on the rotating polygon mirror ll provides six sweep- of each object during one revolution of the rotary the polygon mirror. Due to the rotation, each seen to turn the light beam from a position H furthest to right in Fig. 3 to a position V on the far left in fig 3.

Next, reference is made to Fig. 5. The figure is illustrated how the object 1 is swept by the incident light beam.

The surface 2 is a path on which the object 1 lies and moves moving forward. The incident light beam moves either across the course and perpendicular to the course plane, as shown at A or parallel to the path flat and perpendicular to the surface 16, as shown at B.

Depending on the speed of the web or the sweep for the incident light beam, the object is thus divided 1 in "discs".

At A, the object is divided from above into a number of adjacent discs, with a viewer thus from above sees the thickness and length of each disc, ie sees the disc in contour, but do not see how high the disc is. At B is divided object 2 from the side in slices. A viewer can from side see the contour of the object, but can not form a direct idea of the width of the object.

In the preferred embodiment of the present invention, the object 2 is divided in this way, wherein the division takes place simultaneously from above according to A and from the side according to B.

The reflected light beam, which is the result of the sweep with the incident light beam, is considered according to Fig. 3 from a location at the light source. The one shown in Fig. 4 10 l5 20 25 30 35 462 696 9 the intensity curve thus corresponds to a "slice" of the present the goal and the course taken from one direction. Several such intentions density curves, which correspond to several consecutive swept, gives an overall picture of the object.

Because the same object is successively swept from two different ones hold, a viewer obtains an approximate image in three dimensions of the surface 2 and the object 1. It will be appreciated however, that the image is not a true three-dimensional image.

Next, reference is made to Figs. 6a and 6b. When used of the preferred embodiment of the invention the detector 9 that for each facet of the rotating polygon the mirror 11 to produce a curve similar to that of FIG 6a. The digitized equivalent is shown in Fig. 6b.

The reference numeral 24 indicates the intensity curve obtained when the light beam is swept in the vertical direction from and down over the right object 1 in Fig. 3. Reference the designation 25 shows the intensity curve occurs when the light beam is swept over the object l from the right to the left in Fig. 3. The reference numeral 26 indicates it intensity curve produced when the light beam is swept over the left object l from right to left. Hän- the display designation 27 indicates the intensity curve is obtained when the light beam is swept in the vertical direction from and up over the left object 1 in Fig. 3.

The interval 24 starts on the left with the high white level 22, which is obtained for the first time as the light beam hits the surface 16. When the light beam has passed a certain distance 28, the object 1 and the intensity are found drops drastically to the black level 21. In the Shown the embodiment increases the intensity as the light beam sweeps over the object before the intensity again decreases to the black level. The distance between the points where the black level is obtained denotes the projected object height. After the object has been swept, it is found again the surface 16, unless the object rests completely against the surface 2 and thereby obscures the surface 16. 462 696 10 15 20 25 30 35 10 Between the interval 24 and the following interval There is a distance below which the intensity is undefined, because the incident light beam does not is focused on any surface 2 or object 1. It will be appreciated that the curve shown in Fig. 6a corresponds to an ideal cases, without disturbances. Then take the interval 25 which includes the width of the object. Before the light beam moves over to the second, left object in the passage the light beam is in a position where the reflected beam Reflected directly back by the rotating polygon mirror ll.

Of course, the above reasoning also applies the analog intensity signals 26 and 27.

In computer processing, it is preferably suitable to use digital signals. A digital repre- sentation of the upper analog signal is effected on known wise with a bit resolution, using a reference level 40 with all signals below this level are "low" and all signals above this level are "high".

Next, Fig. 7 is shown. Fig. 7 shows a block diagram describing the main components of the signal processing in accordance with the invention.

The light that has been reflected is perceived by a diode 100 in the detector 9. The analog output signal from the diode 100 is similar to the intensity representing signal len in Fig. 6a. The signal passes an amplifier 101 with frequency-dependent filters, for filtering unwanted frequencies venser. The filtered and amplified signal is fed to a black level latch 102. This latch circuit sees the signal at the lowest intensity, which the black level is mentioned. The signal is sent to a constraint circuit 103 for the white level, which circuit limits the signal maximum level, and then on to a comparator 104, which also receives a reference signal from a computer 113.

Using the reference signal, the comparator 104 generates as above a digital pulse train with ones and zeros of the signal. 10 15 20 25 30 35 462 696 ll This digital signal is fed to two logic controlled shift registers 106 and 107. A pattern detector 108 resp 109 is connected to each shift register l06 and 107, respectively, and these components together form two parallel ones sequence networks 106, 108 and 107, 109. l06, 108 is fed with the signal from the comparator, while One sequence network the second sequence network 107, 109 is fed with the inverse the signal. The task of the pattern detectors is to identify pattern of certain predetermined pulse train sequences. Pattern- the detectors only output these patterns identified in the signal. Figures 6a and 6b show at 25 and 26 how two different sequences of pulses can be center was each object. At 25 gives the front edge of the object gives rise to a pulse trailing edge 50, and the trailing edge of the object gives rise to a pulse leading edge 51. At 26 gives the object leading edge giving rise to a pulse trailing edge 60, and the object's trailing edge giving rise to a pulse leading edge 63. In addition, the object's good reflectivity an intensity that locally exceeds the reference level 40. Thereby obtained another pulse having a pulse leading edge 61 and a pulse back 62. p Then one of the pattern detectors gives an output signal is fed to a buffer memory 112. The buffer memory is also fed with a time value from a time base ll0, which is triggered by a start sync signal from a photodiode lll. The photodiode III is affected by light once per phase. seen on the rotating polygon mirror. It means that the time base is reset for each facet, ie for each measurement cycle that includes scanning two objects from two different directions.

The time value corresponds to the specific angle it the rotating polygon mirror occupies for a certain leading edge or trailing edge of the object. These numbers are fed through the buffer memory in a queue.

The computer ll3 retrieves the numbers from the buffer memory needed to determine the objects. The computer 113 is connected to an output device 114 which may be a display 462 696 10 15 20 25 30 35 12 or similar. In addition, the computer is connected to one sorting means 115, which is supplied with a control signal from the computer for performing sorting functions.

The computer determines by means of the contents of the buffer memory the numbers understood which numbers belong to which object. IN the preferred embodiment, in which two objects while being scanned, the objects' data is of course shared into two separate registers. From now on, the handling comes of a register to be described, it being understood that both the registers are treated in the same way.

The next step for the computer is to combine the developed ones related numbers to distances denoting the object width and height and thereby form "discs", whereby the disc thickness is determined by the distance between each sweep, such as shown in Fig. 5. The distance will of course depend on the band speed and the rotation of the rotating polygon mirror speed.

These straight block-like discs are superimposed, ie lined up one after the other, with a body approximated actively mimics the object formed. If the object example- is a potato, the volume of the potato is of course re than this rectangular body. The developed the volume of this approximate body is multiplied preferably with a form factor specific to each type of object. For a potato that looks like one ellipsoid, it may be appropriate to multiply the volume with a form factor of about 80% for an imitation of a correct volume.

Even after multiplication by a calibration, the computer can row density factor calculate the weight of the objects. Potato density can vary very significantly and it is appropriate that in a known manner on the basis of a number of samples liberate the density factor.

The computer further calculates the maximum projected length. the one, which may differ from the length a viewer looks from the side of the band, in accordance with the discussion around fig la-lc. The computer then also calculates the maximum 10 15 20 25 30 35 462 696 13 projected width that is perpendicular to the length as well the maximum projected height from the side.

Consumers place very different demands on, for example potatoes. Restaurant kitchens want small and round, while food Manufacturers may find it desirable to get long and thick potatoes. In this case, it is appropriate to the computer calculates a "eccentricity factor" which to some extent responds to the visual experience the consumer gets. This is a feature which distinguishes the present invention the most significantly from those on the market for potatoes intended mechanical sorting devices, in which e.g. sieve stitches are used. The problems with these devices are mentioned des in the introduction.

The computer also calculates where the object is located on the bank. that is, ie where the object begins and ends seen in the band longitudinal direction, and where the object begins and ends seen in the transverse direction of the belt. In this case, a geometric trical mean as for an ideal, symmetrical body would correspond to the center of mass. These values are used for calculation of control signals for a sorting means 115.

Furthermore, the computer adjusts the sorting to apply for example, screen mesh sorting to correspond to the criteria laid down for potatoes. Nature- of course, any other sorting specification fication that corresponds to certain individual requirements is entered in the computer.

The preferred scanning device according to the present invention The present invention is summarized in the following way. A batch of potatoes is loaded on two parallel runs bands, which are guided under the scanning device. The the rotating polygon mirror sweeps a laser beam over the potatoes, so that the potatoes and the strips are wrapped one once for every millimeter on the belt. It reflected the light is perceived by the detector which converts the intensity to computer processable signals. The computer emits control signals to the sorting device mounted downstream of the belt, after which the potatoes can be sorted 462 696 10 15 20 25 14 out according to appropriate criteria in certain lots.

Simultaneous with the measurement of size, shape etc can the object is also identified. With regard to it for each object-specific reflection curve, a goals are identified. It is common knowledge that for example potatoes absorb at a light wavelength of 1400 nm while, for example, stone, which in appearance may resemble potatoes, reflects at this wavelength.

The identification of objects can be accomplished on several ways. In the preferred embodiment, i the light-transmitting means 8 an additional lamp is provided. nad. The reflected light is divided into two different waves. longitudinal band and detected. The intensity of each wave longitudinal bands are compared with the reflection spectrum. In it simplest embodiment divides the intensities and in case of exceeding or falling below a certain value of the quota indicates the identity of the object.

The computer compares the object's shape and identity. One number of different selection criteria then governs the generation of the control signal to the sorting device. In the lecture In the embodiment, objects are sorted that are not specific predetermined security can be determined to be potatoes away. Of course, one and the same light source, such as one halogen lamp, is used in both form determination and identification.

Those skilled in the art can, of course, construct alternatives embodiments, without departing from the invention. even scope of protection.

Claims (7)

10 15 20 25 30 35 462 696 15 PATENT REQUIREMENTS Method for determining an approximate, three-dimensional image of an object (1), which is moved on a reflective transport path (2) past a scanning device (30), characterized in that a narrow light beam from the scanning device is swept partly across the transport path , partly across a surface (16), which is located at one longitudinal side edge of the conveyor track and is perpendicular to the conveyor track; that light reflected back to the scanning device is detected; that the intensity of the reflected light is recorded, whereby high intensity is obtained when the light beam hits the transport path and the surface, respectively, while the intensity decreases significantly when the light beam hits the object, and whereby high intensity is again obtained when the light leaves the object, to form a straight block-like image. of the swept portion of the object and for determining the position of that portion on the conveyor track, which image has dimensions corresponding to the width, height and thickness of the portion; that the images formed by successive sweeps are superimposed to form the three-dimensional image of the object and thereby determine the position of the object on the transport path. 2. A method according to claim 1, wherein two parallel, reflective transport paths which move objects past a common scanning device (30) are swept simultaneously. 3. A method according to any one of the preceding claims, characterized in that the detected light is converted into a digital pulse train with level changes corresponding to the intensity; that the pulse train is compared with a predetermined pattern of pulses so that certain sequences of pulses are converted into blocks, which represent distances between two points on contours on the section of the object; 4 2 6 10 l5 20 25 30 35 6 16 that a plurality of blocks are registered for the approximate image of the object; and that the position of the object on the belt is calculated. 4. A method according to any one of claims 1-3, characterized in that the light from the light source and / or at least a second light source is divided in a manner known per se into at least two wavelength-dependent light fluxes, the intensity of which is converted to signals that are compared with a pre-determined spectrum of reflection for the object, for identification of the object. Device for determining an approximate, three-dimensional image of an object (1), which is moved on a reflective transport path (2) past a scanning device (30), this comprising partly a light-emitting means (8) , which sends light in the form of a narrow light beam towards a distribution means (11, 12, 13), partly a detector means (9), partly an electronics and computer circuit (17), characterized in that the distribution means (11, 12, 13 ) is arranged to divide the light beam into at least two light beams angled to each other for sweeping, partly across the conveyor track (2), partly across a surface (6) which is located at one longitudinal side edge of the conveyor track and is substantially perpendicular towards the transport path; that the detector means (9) is arranged for detecting reflected light, which is reflected back to the scanning device (30), that the electronics and computer circuit (17) are connected to the detector means (9) and arranged to process a signal corresponding to the reflected light. detected intensity for obtaining the approximate, three-dimensional image and calculating the position of the object on the transport path (2). Device according to claim 5, characterized in that the distributing means (11, 12, 13) is arranged to divide the narrow light beam into two light beams for sweeping over two parallel transport paths. (2, 2) - Device according to claim 6, characterized in that the distributing means (11, 12, 13) has a rotating polygon mirror (11), which transmits the light beam symmetrically in two opposite directions so that each of the two different objects (1) is swept alternately and in the same way.