JP2003083718A - Visual inspection system for high-speed moving body - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a visual inspection system by which the abnormality of the outside shape of many moving bodies conveyed by a conveyor at a high speed and the abnormality of a position on the conveyor are detected at the high speed so as to issue a warning. SOLUTION: Molded products 30 on the conveyor 10 are monitored by a surface camera 20 and side-face cameras 21 to 24. The surface camera 20 monitors the surface shape of each molded product 30 and an abnormality regarding the position on the conveyor 10. The cameras 21 to 24 obtain information on the passage position of each molded product 30 by the surface camera 20, they share the monitoring of an abnormality regarding the side-face shape of each molded product 20 so as to detect the abnormality jointly. When the abnormality is detected, an abnormality warning display 33 as an upstream countermeasure output and a dumping nozzle 31 as a downstream countermeasure output are driven.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a large number of molded products in which, for example, cylindrical bottles molded by a plurality of molding machines are mixed and conveyed by a single line of conveyor, and the appearance shape and the position on the conveyor are abnormal. High-speed moving object visual inspection system for detecting objects, etc., especially multiple CCs working together
The present invention relates to an improved appearance inspection system for a high-speed moving body that enables high-speed processing by a D camera.

[0002]

2. Description of the Related Art For articles moving on a conveyor, 1
It is widely practiced to perform visual inspection using one or a plurality of CCD cameras.

Japanese Patent Laid-Open No. 2000-337843 discloses that
Regarding a minute chip-shaped electronic component that moves on a belt conveyor, an appearance inspection device using four cameras for monitoring the upper and lower surfaces and the left and right side surfaces is described. However, in this appearance inspection device, the four surfaces are simply shared and monitored by four cameras, and information is not mutually exchanged.

Further, as for a method of processing an image taken by a CCD camera, various methods are utilized in appropriate places.

In Japanese Patent Laid-Open No. 2001-126069,
Described is an improved image recognition method for detecting the position of a detection target by performing pattern matching so that a contour image obtained by photographing a recognition target and a contour image previously learned and stored are the same contour image. ing. However, this image recognition method does not relate to a moving object, but merely describes an efficient comparison / match determination method for a pair of contour images.

On the other hand, in the bottle making process, the contour of the bottle is detected, the inner surface of the bottle is detected, the appearance deterioration of the coated bottle is detected,
CCD cameras are used for detecting defects on the top surface of the aperture, detecting uneven thickness, detecting surface color, and the like.

[0007] In Japanese Patent Laid-Open No. 2000-55831, a CCD camera is used to read a die number code provided on the bottom surface of a bottle for a bottle produced by a die having a plurality of sections, and the bottle is used at various places in the production line. A bottle inspection system using a network system that aggregates and manages inspection data by a plurality of provided bottle inspection devices is described.

Further, Japanese Patent Publication No. 8-506558 describes in detail a glass product conveying mechanism in which high-heat glass bottles manufactured by a mold having a plurality of sections are sequentially arranged and mounted on a conveyor. However, since the glass bottles are not fixed on the conveyer, the arrangement position and the mounting interval of the glass bottles are uncertain, and a fall bottle may occur. Especially, when the position of the bottle loaded from a specific section is abnormal, it is necessary to check the bottle loading mechanism. Furthermore, when the external shape of the bottle loaded from a specific section is abnormal, it is necessary to replace the mold of the section.

In addition, in Japanese Patent Laid-Open No. 157523/1993, a hot glass is used to detect the shape of a glass bottle immediately after molding by a shutter camera installed above and to the side of a conveyor and to detect a bottle having a large degree of deformation by image processing. A method for inspecting the shape of a glass bottle at the end is described.
The present invention is a high-speed inspection method at the hot end using a plurality of cameras, but the upper and side cameras merely share the monitoring surface, and they do not mutually perform function interpolation.

[0010]

In the conventional appearance inspection system as described above, since bottles and the like immediately after molding are conveyed at high speed at unspecified positions on the conveyor, high-speed and accurate appearance shape inspection can be performed. It is difficult to read the die code number, and it is common to inspect the bottle in a post-process where the bottle is cooled and the inspection becomes easy. In this case, since the detection of mold abnormality will be delayed, a large number of defective bottles will be manufactured, and exception management will be required to melt and mold again, and the operating efficiency of the equipment will also decrease. was there.

The present invention has been made to solve the above-mentioned problems, and in addition to detecting an abnormality in the position on the conveyor of a moving body which is transported at a high speed at an unspecified position on the conveyor, it also causes an abnormality in the external shape. It is an object to obtain a visual inspection system for a high-speed moving object that enables detection.

Further, the present invention has been made to solve the above-mentioned problems, and does not directly read the die code number provided on the high-speed moving body, but uses an indirect method to perform the die code number. It is an object of the present invention to obtain an appearance inspection system for a high-speed moving body, which can judge the situation and quickly take countermeasures against abnormalities.

[0013]

SUMMARY OF THE INVENTION Claims 1 to 3 of the present invention
In the high-speed moving body appearance inspection system according to the first aspect, the moving body that is conveyed at high speed by the conveyor is monitored by the first camera for monitoring the upper surface and the second camera for monitoring the side surface. Moreover, the first camera monitors the top shape of the moving body for abnormalities, and the position information on the conveyor is obtained to determine the shooting timing of the second camera. I do. This is because the first and second cameras do not simply share different monitoring surfaces, but effectively utilize the position information of the moving body that moves at a high speed at an unspecified position on the conveyor to improve the cooperation function of both cameras. It provides a basic means of doing.

In the appearance inspection system for a high-speed moving body according to claim 4 of the present invention, the number of second cameras for side surface monitoring is increased to perform multi-direction side surface monitoring of the high-speed moving object, thereby achieving high speed, The present invention provides a means for highly accurately determining a shape abnormality.

In the visual inspection system for a high-speed moving object according to claim 5 of the present invention, the second camera for side surface monitoring is a low-cost because the moving object is photographed at a plurality of timings to carry out side surface monitoring in multiple directions. A means for highly accurately determining a shape abnormality is provided.

In the visual inspection system for a high-speed moving body according to claim 6 of the present invention, the first camera for upper surface monitoring is installed at the upstream position of the conveyor as compared with the second camera for side surface monitoring, and the same moving body is installed. On the other hand, after the first camera captures the top image, the second camera captures the side image when the moving body moves to the position of the second camera. This provides a means for obtaining a time margin for calculating and calculating the timing of side face photography by the second camera after obtaining the position information of the moving body by the first camera.

In the visual inspection system for a high-speed moving body according to claim 7 of the present invention, the first camera for monitoring the upper surface is configured to photograph the upper surface of the moving body at a predetermined time or at each predetermined moving amount of the conveyor. There is. This is to provide a means for avoiding omission of photographing by always performing top-side photographing once or twice when the high-speed moving body is within the field of view of the first camera.

In the visual inspection system for a high-speed moving body according to claim 8 of the present invention, the second side monitoring camera measures the required moving distance from the time when the first camera shoots the top surface of the moving body to the side shooting position. It is configured to capture a side view of the moving body at a time point. This provides a means for facilitating the analysis of the side image by controlling the photographing timing by the second camera.

In the visual inspection system for a high-speed moving body according to claim 9 of the present invention, the abnormality detection of the upper surface shape by the first camera is performed by comparing the learned memory information and the actual driving information. The learning information is learned and stored as secondary upper surface contour image information which is interpolated and has center points at various fixed point coordinates from primary upper surface contour image information having unspecified center point coordinates during learning operation. Further, the upper surface contour image information having unspecified actual measurement center point coordinates during practical operation is the learning read image information obtained by interpolating the secondary upper surface outline image information at the adjacent fixed point coordinates that are learned and stored into the actual measurement center point coordinates. It is configured to compare between. The above-mentioned top surface contour image information is, for example, the simplest to grasp as the width dimension of the contour image in the two axial directions, and this width dimension is the diameter of the cylinder if the moving body is a cylinder and is photographed from the point at infinity immediately above. It represents itself. However, the positional relationship between the first camera and the moving body is unspecified, and the biaxial dimension of the top contour image changes depending on the height of the moving body and the relative position with respect to the first camera. Therefore, the learning information is replaced with a large number of fixed point coordinate position information and stored, and this stored information is reconverted to the center point coordinate position of the moving body actually measured in practical operation and used as comparison reference information. Means are taken.

In the appearance inspection system for a high-speed moving body according to claim 10 of the present invention, the abnormality detection of the upper surface shape by the first camera is performed by comparing the learned memory information with the actual driving information similarly to the above-mentioned claim 7. . However, as the first camera for upper surface monitoring, a plurality of cameras installed at mutually different positions are used, and the upper surface contour image information at the adjacent fixed point coordinates is immediately calculated by interpolation calculation from the plurality of upper surface contour images by the plurality of cameras. The average of many calculated information is learned and stored. Further, even in the practical operation stage, since the upper surface contour image information at the adjacent fixed point coordinates can be immediately obtained from the plural upper surface contour images by the plural cameras by the interpolation calculation, the measurement conversion information and the learning memory information are compared to detect the abnormality of the upper surface shape. It is configured so that it can be judged. This provides a second means for top surface comparison.

In the appearance inspection system for a high-speed moving body according to claim 11 of the present invention, the abnormality detection of the side surface shape by the plurality of second cameras is performed by comparing the learning memory information and the actual driving information as in the case of the upper surface shape. Done. However, in the side image capturing of the moving object by the second camera, the image capturing timing is controlled by using the position information of the moving object by the first camera, so that the image capturing at the position far from the predetermined common coordinate line is not performed. By doing so, the conversion process of the image information can be approximately and easily performed. In addition, the learning information is stored as one piece of averaged image information obtained by averaging the center points of a large number of photographed side surface contour images that have been moved and transformed on a predetermined common coordinate line. Further, the center point of the side surface contour image photographed in the practical operation stage is also moved and converted onto the predetermined common coordinate line, and the side surface shape abnormality is determined by comparing the converted image information with the averaged image information. It is like this. This is for the problem that the side shape of the moving body changes depending on the passing position of the moving body.
The present invention provides a means for efficiently determining an abnormality in a side surface shape by integrally managing each contour image as information obtained by moving and converting each contour image on a predetermined common coordinate line.

In the visual inspection system for a high-speed moving body according to claim 12 of the present invention, the abnormality of the side surface shape detected by the second camera is detected by comparing the learned memory information with the actual driving information as in the case of claim 9. Be seen. However, the second camera performs side image capturing twice on one moving body at different positions, and the predetermined common coordinate line is set at two positions corresponding to the different positions. This provides a means for obtaining side image information at more viewing angle positions from one second camera.

In the appearance inspection system for a high-speed moving body according to claim 13 of the present invention, the abnormality detection of the side surface shape by the second camera is carried out without depending on the learning memory information. A plurality of second side-face monitoring cameras are used, and the side-face contour image information obtained by actual measurement is compared with each other, and if they match, it is determined to be normal. However, the center point of the side face contour image information to be compared is the transformed contour image information after moving on the predetermined common coordinate line. This is to provide a side face abnormality detecting means that does not require learning and memory by converting side face contour images from a plurality of cameras having different viewpoint positions as common information that can be compared with each other.

In the appearance inspection system for a high-speed moving body according to claim 14 of the present invention, the high-speed moving body conveyed by the conveyor is a molded product molded by a molding machine having a plurality of sections, and a mounting signal of the molded product is provided. And tracking means for moving the section number or die number of the applied molding machine based on the conveyor movement signal. This provides a means for reading and displaying the section number or mold number of the molding machine that molded the abnormally detected molded product without the need to read the mold number or the like provided on the bottom surface or side surface of the molded product. It was done.

In the visual inspection system for a high-speed moving body according to claim 15 of the present invention, when abnormal position of the moving body on the conveyor or abnormal shape of the upper surface or side surface is detected, discharge processing of the moving body as a downstream countermeasure output is performed. Then, the molding machine section number, the mold number, or the abnormal classification information as the upstream countermeasure output is read and displayed. This provides a means to quickly take upstream measures by identifying which mold is abnormal or which section mounting mechanism is abnormal, in addition to the discharge processing of the moving body itself in which an abnormality has occurred. It was done.

In the visual inspection system for a high-speed moving body according to claim 16 of the present invention, the first and second cameras for monitoring the moving body are housed in an air-cooled box for supplying and exhausting cooling air,
The molded product, which is a moving body, is photographed through a heat-resistant transparent glass provided on the wall surface of the air-cooled box. This provides a means for enabling close-up photography of a high-temperature moving body and obtaining a clear image.

[0027]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiment 1. A visual inspection system for a high-speed moving body according to Embodiment 1 of the present invention will be described with reference to the drawings. 1 is a diagram showing a front configuration of a visual inspection system for a high-speed moving body according to a first embodiment of the present invention. In each figure, the same reference numerals indicate the same or corresponding parts.

In FIG. 1, 1a to 9a are molding machines for multiple sections using molds of the same shape, and 10 is a motor 11.
Is a conveyor driven in the direction of arrow 13 and reference numeral 12 is an encoder for measuring the amount of movement of the conveyor 10. The molded product 30 formed by each of the molding machines 1a to 9a is mounted on the conveyor 10 by a mounting mechanism (not shown) and moves in the direction of arrow 13 as a large number of moving bodies.

Further, in the figure, 18 is a conveyor 10
Is a horizontal conveyor configured to change the flow of the molded product 30 in the direction of the arrow 17 and 19 is a slow cooling stocker in which the molded products 30 are sequentially stored from the horizontal conveyor 18. Molded product 3 cooled in this slow cooling stocker 19
0 is packed by an appropriate quantity by a packing mechanism (not shown).

Further, in the figure, reference numeral 20 denotes an upper surface camera (first camera) for photographing the upper surface of the moving body 30 which is a molded product, and the upper surface camera 20 is on the moving center line of the conveyor 10 and moves the moving body 30. Although it is installed at a position slightly higher than the height of the moving body 30, the position and the interval of the moving body 30 are not necessarily constant.

In the figure, reference numerals 21 to 24 denote four first to fourth side cameras (second cameras) for photographing the first to fourth side surfaces of the moving body 30.
˜24 are installed at a position downstream of the top camera 20 and facing a predetermined common coordinate line standing at one point on the moving center line of the conveyor 10.

Further, in the figure, 31 is a first to a fourth.
A waste nozzle (downstream countermeasure output means) that is installed at a position downstream of the side cameras 21 to 24 and blows an air flow to the molded product 30 moving on the conveyor 10 to remove the molded product 30 from the conveyor 10. , 32A
Is an appearance inspection device mainly composed of a microprocessor described later, and 33 is an abnormality alarm indicator (upstream countermeasure output means) that displays the content when the appearance inspection device 32A detects an abnormality relating to the position or shape of the molded product 30. is there. The appearance inspection device 32A also operates the waste nozzle 31 when an abnormality relating to the position or shape of the molded product 30 is detected.

FIG. 2 is a diagram showing the structure of the appearance inspection apparatus of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

In FIG. 2, 200A is a visual inspection device 3.
2A is the first inspection unit, and the first inspection unit 200A is
A microprocessor (CPU: abnormality detection means) 40a,
System memory (ROM) 41a, arithmetic memory (RA
M) 42a.

In the figure, reference numeral 34 is a learning switch which is turned on during the learning operation.
4 is input to the first inspection unit 200A and a second inspection unit described later.

Further, in the figure, reference numeral 20a denotes a shutter (ST) of the top camera 20, and this shutter 20a
Are driven by the first inspection unit 200A. A captured image signal from the top camera 20 and a movement signal of the conveyor 10 from the encoder (EC) 12 are input to the first inspection unit 200A, and the discard nozzle (VB) 31 is input.
And a display signal for the abnormality alarm indicator 33 are output.

In the figure, reference numeral 300A denotes a second inspection section of the visual inspection apparatus 32A.
A is a microprocessor (CPU: abnormality detection means) 4
0b, a system memory (ROM) 41b, and an arithmetic memory (RAM) 42b.

Further, in the figure, 21a to 24a are shutters (ST) of the first to fourth side cameras 21 to 24, and these shutters 21a to 24a are driven by the second inspection section 300A. Has become. Also,
The second inspection unit 300A includes the first to fourth side cameras 21 to 21.
A photographed image signal by 24, a movement signal of the conveyor 10 by the encoder (EC) 12, and an operation mode signal by the learning switch 34 are input.

Next, the operation of the appearance inspection system for a high-speed moving body according to the first embodiment will be described with reference to the drawings.

FIG. 3 is a field-of-view plan view of the top camera of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

In FIG. 3A, the top camera 20 is
It is installed right above the coordinate origin P1a on the moving center line 35 of the conveyor 10. The molded product 30 has a bottom position deviation L
Although it is moving to the left on the moving line 36a having 1a, the true value of the position deviation L1a is unknown (cannot be photographed by the camera), and the correction coefficient K is estimated by the procedure described later to calculate the top contour image. It is calculated by multiplying the value of the center point coordinate by the value of the correction coefficient.

That is, the coordinates of the center point P2 of the top surface contour image of the molded product 30 are shown as (L2, L1) with respect to the origin P1a, but the coordinates of the bottom center point P2a of the molded product 30 are (L2a, L1a), L2a≈K × L2, L
It is assumed that 1a≈K × L1. However, K (≦ 1.
0) is a correction coefficient, which is a substantially constant value depending on the height and diameter of the molded product 30.

Note that D2 is the width dimension in the moving line direction (x-axis direction) in the upper surface contour image of the molded product 30, and D1 is the width dimension in the direction orthogonal to the moving line (y-axis direction). When the molded product 30 has a cylindrical shape and is moving just below the top camera 20, the dimension is D1 = D2. (D
2 × D1) is referred to as the top contour image size.

P3a is on the conveyor center line 35,
It is a reference point for side surface photography located downstream of the origin P1a by an interval L0 which is an average mounting interval dimension of the molded product 30, and a right-angle normal line on the side surface reference point is the common coordinate line 37b.
As shown.

P4a is a side surface photographing position where the side surface photographing of the molded product 30 is performed by the first side surface camera 21 and the third side surface camera 23, and L3 is the first and third sides after the top surface photographing is performed by the top surface camera 20. Required movement distance until the side cameras 21 and 23 perform side photographing, 37a is side photographing position P
4a is a right-angled normal. As will be described later with reference to FIG. 5, the first and third side surface cameras 21 and 23 are above the side surface camera center line 39a connecting the points P3a and P4a, and have a height corresponding to the average height dimension of the molded product 30. It is installed in the position.

FIG. 3 (b) is a partially enlarged view of FIG. 3 (a). In FIG. 2B, P2 (L2, L1) is the center point of the top contour image when the molded product 30 is photographed by the top camera 20, and P2a (L2a, L1a) is the center of the bottom surface of the molded product 30 at this time. Points, P2b and P2c are correction factors K2
Alternatively, it is the estimated center point of the bottom surface of the molded product 30 obtained by multiplying K1.

On the other hand, the plane coordinates of the side surface image are constituted by the X axis perpendicular to the camera center line 39a connecting the first and third side surface cameras 21 and 23 with the normal line 37b of the side surface reference point P3a as the Y axis. The origin is at the height position of the first and third side cameras 21, 23 on the Y axis. However, in FIG. 3, the height of the side camera is drawn as 0 for convenience.

Here, if the correction coefficient is K2 and it is estimated that the bottom center point of the molded product 30 is at the position P2b, the first,
The required moving distance until the side face is photographed by the third side face cameras 21 and 23 is the distance between the points P2b and B. However, B is the intersection of the movement line of the molded product 30 which is on the point P2b and the center line 39a of the camera. However, since the actual bottom surface center point was at the position of P2a, the bottom surface center point position at the time of the side surface photography was set at the same moving distance from P2a.
It becomes 4ab. Therefore, when this is photographed by the third side camera 23, there is a photographing point error of ΔXb in the X-axis direction.

Similarly, if the correction coefficient is set to K1 and it is estimated that the center point of the bottom surface of the molded product 30 is at the position P2c, the first,
The required movement distance until the side face is photographed by the third side face cameras 21 and 23 is the distance between the points P2c and C. However, C is the intersection of the movement line of the molded product 30 which is on the point P2c and the camera center line 39a. However, since the actual bottom surface center point was at the position of P2a, the bottom surface center point position at the time of the side surface photography was set at the same moving distance from P2a.
4ac. Therefore, when this is photographed by the third side camera 23, there is a photographing point error of ΔXc in the X-axis direction.

In this way, if the correction coefficient K is not appropriate, a shooting point error ΔX will occur, but an appropriate correction coefficient K is sequentially obtained while performing a learning operation as will be described later. .

FIG. 4 is a diagram showing fixed point coordinates and interpolation straight lines of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

In FIG. 4A, P1a is the coordinate origin on the center line of the conveyor immediately below the top camera 20, as described above, while P10a to P17a are peripheral fixed points surrounding the origin P1a, and these peripheral fixed points. P10a-P1
The coordinates of 7a are distributed and set at substantially uniform positions within the field of view of the top camera 20.

Further, in FIG. 7A, P20 and P2
Reference numeral 2 is the center point of the contour image when the molded product 30 moves along the bottom surface movement line 36b and the top surface camera 20 performs two top surface photographing.

Similarly, in FIG.
1 and P23 are the center points of the contour image when the molded product 30 moves on the bottom movement line 36c and the upper surface camera 20 performs two upper surface photographing and shooting.

Here, when the coordinates of the origin P1a are (0, 0), the coordinates of the center point P20 are (x20, y20), and the size of the captured contour image (see D2 × D1 in FIG. 3) is ( D220, D120) and the coordinates of the center point P21 are (x2
1, y21), and the size of the captured contour image is (D22
1, D121), the coordinates are (x16, y16)
The contour image size on the adjacent fixed point P16a (D216,
D116) is calculated by the interpolation calculation as follows.

FIG. 4B is an x-axis direction interpolation straight line diagram, which is for performing interpolation calculation of the x-axis direction dimension.
In the same figure (b), the dimension D220 at the position x20
And the dimension D221 at the position x21 are interpolated to calculate the dimension D216 at the position x16 on the interpolation line 38a by the proportional calculation.

Similarly, FIG. 4C is a linear interpolation diagram in the y-axis direction, which is for performing interpolation calculation of the dimension in the y-axis direction. In the figure (c), the dimension D at the position y20
120 and the dimension D121 at the position y21 are interpolated to obtain the dimension D11 at the position y16 on the interpolation line 38b.
6 is calculated by proportional calculation.

FIG. 4D shows an example of coordinate points when the contour dimension information at the fixed point coordinates that have been learned and stored is read out and used. In FIG. 6D, P22 is the center point of the upper surface contour image of the molded product 30 taken in the practical operation stage. In this case, the contour dimension as the determination criterion at the point P22 is calculated by interpolating the contour dimensions stored in the neighboring fixed points P13a and P16a at the position of the point P22, and the comparison determination is performed using this learning read information as the reference value. It is supposed to be done.

FIG. 5 is a diagram showing the layout and side surfaces of the side camera of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

In FIG. 5A, a pair of first and third side cameras 21, 23 and second and fourth side cameras 22, 24.
Is installed above the position facing the side surface reference point P3a on the conveyor center line 35.

The molded product 30 is moving on the moving line 36a having the bottom position deviation L1a, and the side surface photographing position P4a.
When the camera comes to the top, the first and third side cameras 21 and 23 take images, and when the camera further moves and comes to the second side image taking position P4b, the images are taken by the second and fourth side cameras 22 and 24. Done.

The side contour image of the molded product 30 shown in FIG. 5 (a) is seen from the direction of arrow 39c. The center point P5 of the side contour image is obtained as the intersection of the diagonal lines of the circumscribed quadrangle of the contour image. To be

FIG. 5 (b) is a partially enlarged view of FIG. 5 (a). In the same figure (b), R is the distance between the side surface photographing position P4a on the moving line 36a and the third side surface camera 23,
ΔR is the distance between the side surface shooting position P4a and the side surface reference point P3a, R / (R + ΔR) is the reduction ratio, and (R + ΔR) / R
Is called the enlargement ratio.

As is apparent from FIG. 5B, the plane body S34 obtained by converting the side face contour image into the side plane reference point P3a when the plane body S4a on the side face photographing position P4a is seen from the third side camera 23 is Coplanar body S3 on the side surface reference point P3a
Since it is enlarged by the difference in the viewpoint distance as compared with the side face contour image when a is viewed, the equivalent side face contour image information on the side face reference point P3a can be obtained by performing back calculation reduction.

Similarly, the side profile image when the molded product 30 on the side surface photographing position P4a is viewed from the first side camera 21 is the side profile image when the same molded product 30 on the side surface reference point P3a is viewed. Compared with this, the distance is reduced by the difference in the viewpoint distance.
The equivalent side contour image information on 3a will be obtained.

FIG. 6 is a flow chart showing the operation of the microprocessor of the first inspection section of the appearance inspection apparatus of the appearance inspection system for high-speed moving bodies according to the first embodiment of the present invention.

In step 200 of FIG. 6, the microprocessor 40a is activated irrespective of the position of the molded product 30 at regular time intervals or whenever the conveyor 10 moves a predetermined distance. That is, this step 200 is an operation start step.

Next, in step 201, the operation mode is determined based on the operating state of the learning switch 34.

Next, in step 202, when it is determined in step 201 that the learning operation mode is set, the upper surface camera 20 photographs the upper surface of the molded product 30.

Next, in step 203, an upper surface contour image is generated, its x-axis direction dimension and y-axis direction dimension, and the center point coordinates of the contour image are detected and primary learning is stored (primary learning storage means).

Next, at step 204, the correction coefficient K is calculated and corrected by the procedure described later.

Next, at step 205, the value of the bottom center position deviation (L2a, L1a) of FIG. 3 is transmitted to the microprocessor 40b of the second inspection section 300A. The value of the bottom center point position deviation (L2a, L1a) is the center point coordinate (L2, L) of the contour image detected in step 203.
The correction coefficient K calculated in step 204 above to the value of 1)
The initial value of the correction coefficient K is set to, for example, about 0.9.

Next, in step 206, among the many pieces of image information sequentially stored in step 203, those whose center point coordinate positions are very close are sequentially averaged to reduce the number of points in the storage memory.

Then, in step 207, the operation ends. After the operation is completed, the operation is started again after a waiting time.

In step 210, the above step 2
When it is determined in 01 that the learning operation mode is not set, it is determined whether the learning operation is the first practical operation operation completed.

Next, in step 211, when it is determined in step 210 that it has just been learned, various fixed point coordinates P10a are calculated based on the contour image dimensions at a large number of center point coordinates stored in step 206. ~ P1
The contour image size at 7a, the origin P1a, etc. is interpolated and stored as a secondary learning value (secondary learning storage means).
The procedure is as described in FIGS. 4B and 4C.

Next, in step 212, when it is determined in step 210 that it is not just after learning, or following step 211, the upper surface of the molded product 30 in a practical operation state is photographed.

Next, in step 213, an upper surface contour image is generated, and its x-axis direction dimension and y-axis direction dimension and the center point coordinates of the contour image are detected and stored (primary measurement storage means).

Next, at step 214, the detected position deviation L1 (value on the y-axis of the center point of the contour image) is compared with a predetermined limit value.

Next, in step 215, when it is determined in step 214 that the deviation is excessive, a first abnormality detection output is generated (first abnormality detection output means).

On the other hand, when it is determined in step 214 that the position deviation is normal in step 216, the center of the contour image detected in step 213 is interpolated from the coordinate size of the adjacent fixed point coordinates stored in step 211. An interpolated contour image dimension related to point coordinates is calculated (see FIG. 4D) (learning read storage means). The image size calculated in step 216 is treated as a tertiary learning value.

Next, in step 217, the contour image size detected in step 213 and the step 2
The dimensions of the tertiary learning contour image calculated in 16 are compared (comparing means).

Next, in step 218, when it is determined in step 217 that the comparison is normal, the value of the bottom surface center point position deviation (L2a, L1a) in FIG.
00A to the microprocessor 40b. The value of the bottom center point position deviation (L2a, L1a) is the center point coordinate (L2, L) of the contour image detected in step 213.
The correction coefficient K calculated in step 204 above to the value of 1)
It is a product of

On the other hand, when it is determined in step 217 that the comparative deviation is excessive in step 219,
Generates a second abnormality detection output (upper surface shape abnormality output means,
Second abnormality detection output means). Then, the above step 218
Or, following step 219, the above-mentioned end step 20
It is designed to move to 7.

When the operation of the microprocessor 40a shown in FIG. 6 is summarized, the molded product 30 is, for example, about 180 minutes per minute.
Individually, the molded product 30 is moving from the right to the left on the conveyor 10 at an average interval of about 100 mm, and the molded product 30 is viewed from the top camera 2.
The start step 200 is activated so that the top image is taken several times while passing through the field of view of 0, and the processing time of the series of steps is completed within this time.

For example, 180 or more pieces of learning data are obtained during one minute of learning operation, and this learning data is a contour image size having unspecified and irregular center point coordinates,
Not suitable for use as a criterion.

Therefore, when the learning is completed, the stored data is organized by interpolation calculation as contour image dimension data regarding a large number of fixed point coordinates (step 211). However,
In the learning data collection process, when there are a plurality of data whose center point coordinates are very close to each other, it is avoided to store the data as they are, and to store the average value data of both data (step 206). For the contour image to be determined detected in the practical operation stage, the contour image size at the adjacent fixed point coordinate position stored in learning is interpolated to the center point position of the contour image to be determined (step 216), and then the contour image to be determined. (Step 217).

If there is an abnormality in the mechanism for mounting the molded product 30 on the conveyor 10, the first abnormality detection output (step 21
5) occurs.

A second abnormality detection output (step 219) is generated when the molded product 30 has fallen or there is an abnormality in the upper surface shape of the molded product 30 which is a moving body due to a defective mold.

At step 205 and step 218, the contour center point position deviation (L of FIG. 3) corresponding to the center point coordinates of the contour image detected at step 203 and step 213 is calculated.
2, L1) value is multiplied by the correction coefficient K calculated in step 204 to obtain the bottom surface center point position deviation (L2
a, L1a) is inferred and transmitted to the microprocessor 40b.
When the upper surface is imaged once, only the initial information is transmitted by monitoring a flag (not shown).

FIG. 7 is a flow chart showing the operation of the microprocessor of the second inspection section of the appearance inspection apparatus of the appearance inspection system for high-speed moving bodies according to the first embodiment of the present invention.

In step 300 of FIG. 7, the microprocessor 40b starts operation.

Next, at step 301, the operation mode is judged based on the operating state of the learning switch 34.

Next, when it is determined in step 301 that the learning operation mode is set in step 302, it is determined whether the microprocessor 40a has transmitted the position information (L2a, L1a) shown in step 205 of FIG. To do. If the position information has not been transmitted, the above step 3
It is designed to move to 01.

Next, at step 303, when the reception judgment of the position information is made at step 302, whether the required moving distance L3 (see FIGS. 3 and 5) is reached while interrupt counting the pulse signal of the encoder 12 Determine whether The moving distance L3 is calculated by the following equation.
First time: L3 = L0-L2a-L1a · tan θ, second time: L3 = L0-L2a + L1a · tan θ, where
L0 is the distance from the installation origin P1a of the top surface camera 20 to the side surface reference point P3a, and L2a is the molded product 30 at the time of top surface photography.
Distance between the center point of the bottom of the and the origin P1a in the X-axis direction, L1a.
tan θ is a distance between the side surface reference point P3a and the side surface photographing position P4a or P4b in the X-axis direction.

Next, at step 304, when the result at step 303 is YES, the first and third side cameras 21, 23 perform side photographing of the molded product 30 that has reached the side photographing position P4a, and the primary learning value is obtained. Is stored as (primary learning storage means).

Next, in step 305, the photographing point error ΔX, which is the deviation in the X-axis direction between the center point position of the side surface contour image stored in step 304 and the common coordinate line 37b in FIGS. It is sent to the processor 40a.

Next, at step 306, the positional deviation between the side face reference point P3a and the side face photographing position P4a is calculated.

Next, in step 307, the secondary contour image information converted into the side surface reference position P3a is stored (secondary learning storage means).

Next, in step 308, the side face contour images converted in the reference point position obtained in step 307 are successively averaged and stored as tertiary learning storage data (third learning storage means).

Next, in step 309, if step 304 is the first side image taking, the process returns to step 303, and if it is the second side image taking, the end step 319.
Move to. This step 309 is a comparison / determination step. When returning to step 303, it is determined in step 303 whether or not the side surface photographing position P4b (see FIG. 5) has been reached, and in step 304, the molded product 30 is formed by the second and fourth side surface cameras 22 and 24.
Will be taken from the side.

In step 310, the above step 3
When it is determined in 01 that the learning operation mode is not set, it is determined whether the microprocessor 40a has transmitted the position information shown in step 218 of FIG. If the position information has not been transmitted, the process proceeds to step 301.

Next, at step 311, when the reception judgment of the position information is made at step 310, whether the required movement distance L3 (see FIGS. 3 and 5) is reached while interrupt counting the pulse signal of the encoder 12 is made. Determine whether

Next, in step 312, the first and third side surface cameras 21, 23 perform side surface photographing of the molded product 30 which has reached the side surface photographing position P4a.

Next, in step 313, a contour image of the photographed side surface image is generated and primary contour image information is stored (primary actual measurement storage means).

Next, in step 314, the positional deviation between the side surface reference point P3a and the side surface photographing position P4a is calculated.

Next, in step 315, the secondary contour image information converted into the side surface reference position P3a is stored (secondary actual measurement storage means).

Next, in step 316, the contour image stored in step 315 is compared with the contour image learned and stored in step 308 (comparing means).

Next, in step 317, when the comparison deviation is excessive in step 316, the third abnormality detection output is generated (side surface shape abnormality output means, third abnormality detection output means).

Next, in step 318, if the deviation in the determination in step 316 is small, or following step 317, if step 312 is the first side photographing, the process returns to step 311 and the second time is taken. If it is the side image capturing, the process proceeds to end step 319. This step 318 is a comparison / determination step. If the process returns to step 311, step 311
Then, it is determined whether or not the side surface photographing position P4b (see FIG. 5) has been reached, and in step 312, the side surface photographing of the molded product 30 is performed by the second and fourth side surface cameras 22 and 24.

The operation of the microprocessor 40b shown in FIG. 7 can be summarized as follows.
4a and P4b positions (see FIG. 5), and the image captured here is always converted into the side reference point P3a position for handling.

The side photographing positions P4a and P4b are moved by the moving line 3
6a is the intersection of the side camera center lines 39a and 39b,
Since the value of the position deviation L1a is transmitted from the microprocessor 40a, the microprocessor 40b can calculate the position deviation between P3a / P4a or between P3a / P4b (steps 306 and 314) based on this value. Is.

However, when there is no positional deviation L1a or when the positional deviation L1a is small, points P3a, P4a, and P4b are infinitely close to each other, and there is no need to perform two separate photographing operations. In such a case, P is determined by a determination step (not shown).
The first to fourth side surface cameras 21 are located at the midpoint between the points 4a and P4b.
It is also possible to capture the images of 24 to all at once.

When the dimension L2a in FIG. 3 is transmitted from the microprocessors 40a to 40b, the required moving distance L3 can be calculated on the side of the microprocessor 40b, but this calculation is performed on the side of the microprocessor 40a and the result is calculated. L3 may be transmitted to the microprocessor 40b.

The comparison judgment in step 316 is
It will be carried out four times in total corresponding to the first to fourth side cameras 21 to 24, and when the molded product 30 is viewed from four directions,
The third abnormality detection output (step 317) operates regardless of the direction of abnormality.

Further, when the photographing point error ΔX is transmitted from the microprocessor 40b to the microprocessor 40a in step 305,
In step 204 of FIG. 6, the correction coefficient K is corrected, and an appropriate correction coefficient K is calculated by repeating this feedback operation in the initial stage of the learning operation.

Embodiment 2. A visual inspection system for a high-speed moving body according to Embodiment 2 of the present invention will be described with reference to the drawings. FIG. 8 is a diagram showing the configuration of the appearance inspection apparatus of the appearance inspection system for a high-speed moving body according to Embodiment 2 of the present invention.

In FIG. 8, 400B is a visual inspection device 3.
2B is a first inspection unit, and the first inspection unit 400B is
A microprocessor (CPU: abnormality detection means) 40c,
System memory (ROM) 41c, arithmetic memory (RA
M) 42c is mainly configured.

In the figure, 34 is a learning switch which is turned on during the learning operation.
4 is input to the first inspection unit 400B and a second inspection unit described later.

In the figure, 20a is a molded product 3.
0 of the shutter (ST) of the top camera 20, 25a is the second
It is a shutter (ST) of the top camera (second camera) 25, and each of the shutters 20a and 25a is the first inspection unit 400B.
Is driven by.

In addition, in the figure, the first inspection section 400B
A photographed image signal from the upper surface cameras 20 and 25 and a movement signal of the conveyor 10 by an encoder (EC) 12 for measuring the amount of movement of the conveyor are input to the drive signal for driving a waste nozzle (VB) 31 for discharging an abnormal product, A display signal for the abnormality alarm display 33 is output.

Further, in the figure, reference numeral 500B is a second inspection section of the visual inspection apparatus 32B.
B is a microprocessor (CPU: abnormality detection means) 4
0d, a system memory (ROM) 41d, and an arithmetic memory (RAM) 42d.

Further, in the figure, 22a and 23a are shutters of the second and third side cameras 22 and 23, and the respective shutters 22a and 23a are driven by the second inspection section 500B.

Further, in the figure, the second inspection section 500
To B, the photographed image signals by the second and third side cameras 22, 23, the movement signal of the conveyor 10 by the encoder (EC) 12, and the operation mode signal by the learning switch 34 are input.

Next, the operation of the appearance inspection system for a high-speed moving body according to the second embodiment will be described with reference to the drawings.

FIG. 9 is a field-of-view plan view of the top camera of the appearance inspection system for a high-speed moving body according to the second embodiment of the present invention.

In FIG. 9, P1b, P3a, and P3b are fixed points on the conveyor center line 35, and a pair of upper surface cameras (first cameras) 20 for monitoring the upper surface of the molded product 30 are provided at the point symmetrical positions of the fixed point P1b. 2 A top camera (second first camera) 25 is installed.

Further, in the figure, P2a is a molded product 30 when the upper surface photographing by the upper surface cameras 20 and 25 is performed.
Bottom center point, P21 (L21, L11) is the center point coordinates of the contour image of the molded product 30 taken by the top camera 20, and P22 (L22, L12) is the second top camera 25.
The center point coordinates of the contour image of the molded product 30 photographed by the above, and the bottom surface center point P2a is estimated and calculated in a manner described later. The fixed point P3a is a first reference point for side surface imaging, and the reference point is an intermediate point of the monitoring line 39c connecting the second and third side surface cameras 22 and 23 for side surface monitoring.
The fixed point P3b is the second reference point for side face photography, and is the second and third reference points.
When the molded product 30 is photographed by the side cameras 22 and 23, the molded product 30 is monitored by the monitoring lines 39c and 3c that connect the second and third side cameras 22 and 23 to the first reference point P3a and the second reference point P3b.
It is performed at the time of passing over 9d and 39e.

Further, in the figure, when the molded product 30 is moving along the movement line 36d in the direction of the arrow 13, side surface photographing by the second side camera 22 and the third side camera 23 is performed at the first photographing position P4a. Be seen. At the second shooting position P4b, the third side camera 23 shoots, and at the third shooting position P4c, the second side camera 22 shoots.

The side face contour image photographed at the first photographing position P4a is enlarged and reduced and converted and stored in the position of the first reference point P3a. In the case of FIG. 9, the contour image by the second side face camera 22 is The image is enlarged and converted in proportion to the viewpoint distance, and the contour image by the third side camera 23 is reduced and converted in proportion to the viewpoint distance.

The side contour images photographed at the second and third photographing positions are enlarged and reduced and converted and stored at the position of the second reference point P3b. In the case of FIG. 9, the contour obtained by the second side camera 22 is used. The image is enlarged and converted in proportion to the viewpoint distance, and the contour image by the third side camera 23 is reduced and converted in proportion to the viewpoint distance.

FIG. 10 is a flowchart showing the operation of the microprocessor of the first inspection section of the appearance inspection apparatus of the appearance inspection system for high-speed moving bodies according to the second embodiment of the present invention.

In step 400 of FIG. 10, the microprocessor 40c causes the conveyor 10 to rotate at regular intervals.
Is activated every time the predetermined distance is moved regardless of the position of the molded product 30. This step 400 is an operation start step.

Next, at step 401, the operation mode is judged based on the operating state of the learning switch 34.

Next, in step 402, when it is determined in step 401 that the operation mode is the learning operation mode, the pair of upper surface cameras 20 and the second upper surface camera 25 photograph the upper surface of the molded product 30.

Next, in step 403, a pair of upper surface contour images are generated, their x-axis direction dimension, y-axis direction dimension, and the center point coordinates of the contour image are detected and stored as the primary learning value (primary learning storage). means).

Next, at step 404, the correction coefficient K is calculated and corrected in the manner described later.

Next, in step 405, the value of the bottom surface center point position deviation P2a (L2a, L1a) in FIG. 9 is set to the second value.
It transmits to the microprocessor 40d of the inspection unit 500B. The value of the bottom center point position deviation (L2a, L1a) is calculated in step 404 as the average value of the center point coordinates (L21, L11) and (L22, L12) of the pair of contour images detected in step 403. The calculated correction coefficient K is multiplied and calculated by the following formula. L2a = K × (L21 + L22) / 2 L1a = K × (L11 + L12) / 2

Next, in step 406, the contour image size for the adjacent fixed point coordinate position (see FIG. 4) is calculated by interpolation based on the pair of image information stored in step 403, and is used as a secondary learning value. Memorize (secondary learning memory means).

Then, in step 407, the operation ends. After the operation is finished, wait a while again,
The operation shifts to step 400.

In step 410, the above step 4
When it is determined in 01 that the learning operation mode is not set, it is determined whether the learning operation is the first practical operation operation completed.

Next, in step 411, when it is determined in step 410 that it has just been learned, a large number of contour image dimensions at various fixed point coordinates stored in step 406 are averaged and stored as a tertiary learning value. (Third learning storage means).

Next, in step 412, when it is determined in step 410 that it is not immediately after learning, or subsequent to step 411, the upper surface of the molded product 30 in the practical operation state is set to the pair of upper surface cameras 20 and the second upper surface. Take a picture with the camera 25.

Next, at step 413, a pair of upper surface contour images are generated, and the x-axis direction dimension and the y-axis direction dimension thereof and the center point coordinates of the contour image are detected and stored (primary measurement storage means).

Next, in step 414, the average value L1 = (L11 + L12) / 2 (average value of the y-axis values of the center points of the contour image) of the detected position deviation is compared with a predetermined limit value. L11 and L12 are y-axis values in the coordinates of the center point of the upper surface contour image obtained by the upper surface camera 20 and the second upper surface camera 25.

Next, in step 415, when the deviation is judged to be excessive in step 414, the first abnormality detection output is generated (first abnormality detection output means).

On the other hand, when it is determined in step 414 that the position deviation is normal in step 416, interpolation calculation is performed from the coordinate dimensions relating to the pair of center point coordinates stored in step 413 to obtain the contour image dimension relating to the adjacent fixed point coordinates. Calculate and store (secondary measurement storage means).

Next, in step 417, the contour image size stored in step 411 and the step 4
The contour image dimensions stored in 16 are compared (comparing means).

Next, in step 418, when it is determined that the comparison is normal in step 417, the deviation of the center point P2a in FIG. 9 corresponding to the average value of the center point coordinates of the contour image detected in step 413 is calculated. Coordinate dimensions (L2a,
The value of L1a) is transmitted to the microprocessor 40d.
The value of the bottom surface center point position deviation (L2a, L1a) is calculated in step 404 as the average value of the center point coordinates (L21, L11) (L22, L12) of the pair of contour images detected in step 413. The correction coefficient K is multiplied.

On the other hand, in step 419, when it is determined in step 417 that the comparative deviation is excessive,
Generates a second abnormality detection output (upper surface shape abnormality output means,
Second abnormality detection output means). Then, the above step 41
8 or step 419 followed by the above end step 4
It is set to move to 07.

When the operation of the microprocessor 40c shown in FIG. 10 is summarized, the molded product 30 is, for example, about 180 minutes per minute.
While moving from right to left on the conveyor 10 at an average interval of about 100 mm, several times of top surface photographing are performed while the molded product 30 passes through the field of view of the pair of top surface cameras 20 and the second top camera 25. Starting step 400 as performed above
Is activated, and the processing time of a series of steps is completed within this time. For example, 180 or more pairs of learning data can be obtained during one minute of learning operation. This learning data is a contour image dimension having unspecified and irregular center point coordinates, and is used as a criterion for judgment. Not suitable for use.

Therefore, an interpolation operation is performed from the paired information obtained by the pair of cameras, and the contour image size data regarding the adjacent fixed point coordinates is stored (step 406).
At this stage, various image size data are stored even with the same fixed point coordinates. Therefore, this is averaged and arranged so that one image size is stored for one fixed point coordinate (step 411). ). For a pair of contour images to be judged detected in the practical operation stage, this is interpolated to the image size at the adjacent fixed point coordinates (step 416).
Then, it is directly compared with the contour image size of the adjacent fixed point coordinate position which has already been learned and stored. In short, the difference from the first embodiment is that the data of unspecified points is converted into fixed point data and handled by using a pair of data from different viewpoints.

If there is an abnormality in the mechanism for mounting the molded product 30 on the conveyor 10, the first abnormality detection output (step 41
5) occurs. Further, if there is an abnormality in the upper surface shape of the molded product 30 that is a moving body due to the molded product 30 falling or a defective mold, a second abnormality detection output (step 419) is generated.

In steps 405 and 418, the P2a position coordinate dimension (L2 in FIG. 9) relating to the center point coordinates of the contour image detected in steps 403 and 413 is used.
a, L1a) values are transmitted to the microprocessor 40d, but if the same molded product is photographed twice on the upper surface, only the initial information is transmitted by monitoring a flag (not shown). Has become.

FIG. 11 is a flow chart showing the operation of the microprocessor of the second inspection section of the appearance inspection apparatus of the appearance inspection system for high-speed moving bodies according to the second embodiment of the present invention.

In step 500 of FIG. 11, the microprocessor 40d starts its operation.

Next, in step 501, the operation mode is determined based on the operating state of the learning switch 34.

If it is determined in step 501 that the learning operation mode is set in step 502, it is determined whether the microprocessor 40c has transmitted the position information (L2a, L1a) shown in step 405 of FIG. To do. If the position information has not been transmitted, the process proceeds to step 501.

Next, in step 503, when the reception judgment of the position information is made in step 502, the required moving distance L3 (see FIGS. 3 and 9) is reached while counting the pulse signal of the encoder 12 as an interrupt. Determine whether or not. This moving distance L3 is calculated by the following equation. First time: L3 = L0a−L2a, where L0a is the distance from the origin P1b to the side surface reference point P3a, and L2a is the distance between the center point of the bottom surface of the molded product 30 and the origin P1b in the X-axis direction when the top surface is photographed. Second time: L3 = L0b-L2a-L1a · tan θ, Third time: L3 = L0b-L2a + L1a · tan θ, where L0b is the distance from the origin P1b to the side surface reference point P3b, and L2a is the bottom surface of the molded product 30 at the time of top surface photography. The distance between the center point and the origin P1b in the X-axis direction, L1a · tan θ is the distance in the X-axis direction between the side surface reference point P3b and the side surface photographing position P4b or P4c.

Next, in step 504, when the result in step 503 is YES, the side surface photographing position P4a is determined by the second side surface camera 22 and the third side surface camera 23.
The side surface of the molded product 30 that has reached is captured and stored as a primary learning value.

Next, in step 505, the center point position of the side face contour image stored in step 504 and
The imaging point error ΔX, which is the deviation in the X-axis direction from the coordinate line on the first reference point P3a, is calculated and transmitted to the microprocessor 40c.

Next, in step 509, if step 504 is the first or second side photographing, the process returns to step 503, and if it is the third side photographing, the process proceeds to end step 519. This step 509 is a comparison / determination step.

When the process returns to step 503 for the first time, in step 503 the side face photographing position P4b is selected.
(Refer to FIG. 9), it is determined whether or not step 5 has been reached.
In 04, the side surface of the molded product 30 is photographed by the third side camera 23. However, when the process returns to step 503 for the second time, in step 503, the side photographing position P4c
(Refer to FIG. 9), it is determined whether or not step 5 has been reached.
In 04, the side surface of the molded product 30 at the position P4c is photographed by the second side camera 22.

The photographing point error ΔX in step 505 at the time of the first return and the second return is the second reference point P3b.
It is calculated as the X-axis direction deviation between the coordinate line on the upper side (see FIG. 9) and the center point position of the side outline image.

In Step 510, the above Step 5
When it is determined in 01 that the learning operation mode is not set, it is determined whether the microprocessor 40c has transmitted the position information shown in step 418 of FIG. If the position information has not been transmitted, the process proceeds to step 501.

Next, in step 511, when it is determined in step 510 that the position information has been received, it is determined whether or not the required moving distance L3 has been reached while counting the pulse signal of the encoder 12 as an interrupt.

Next, in step 512, the second side camera 22 and the third side camera 23 perform side view photography of the molded product 30 that has reached the side view photography position P4a.

Next, in step 513, a contour image of the photographed side surface image is generated and primary contour image information is stored (primary actual measurement storage means).

Next, in step 514, the positional deviation between the side surface reference point P3a and the side surface photographing position P4a is calculated.

Next, in step 515, the secondary contour image information converted into the side surface reference position P3a is stored (secondary actual measurement storage means).

Next, at step 516, the pair of contour images stored at step 515 are compared (comparing means).

Next, in step 517, when the comparison deviation is excessive in step 516, the third abnormality detection output is generated (side surface shape abnormality output means, third abnormality detection output means).

Next, in step 518, when the determination in step 516 has a small deviation, or following step 517, step 512 is the first or second step.
If it is the second side photographing, return to step 511,
If it is the third side photographing, the process proceeds to the end step 519. This step 518 is a comparison / determination step.

If the process first returns to step 511, then in step 511 the side face photographing position P4b is selected.
(Refer to FIG. 9), it is determined whether or not step 5 has been reached.
In 12, the side surface of the molded product 30 is photographed by the third side camera 23.

When the process returns to the step 511 for the second time, the side photographing position P4c is determined in the step 511.
(Refer to FIG. 9), it is determined whether or not step 5 has been reached.
In 12, the side surface of the molded product 30 is photographed by the second side camera 22.

Furthermore, at the time of the initial return operation, step 5
The comparison operation by 16 is omitted, and the comparison is executed at the time of the second return operation. The comparison operation includes the stored contents of step 515 at the time of first return and step 515 at the time of second return.
The memory contents of are compared.

When the operation of the microprocessor 40d shown in FIG. 11 is summarized, the side image of the molded product 30 is photographed.
At positions P4a, P4b, P4c (see FIG. 9),
The image captured at the point P4a is converted to the side reference point P3a position and handled, and the image captured at the points P4b and P4c is converted to the side reference point P3b position and handled.

Side photographing positions P4a, P4b, P4c are
Moving line 36d and side camera surveillance lines 39c, 39d, 3
9e and the value of the position deviation L1a is transmitted from the microprocessor 40c. Based on this, the microprocessor 40d calculates the position deviation between P3a / P4a or between P3b / P4b and P4c (step 514). ) Is possible. However, when the position deviation L1a does not exist or is small, P3b, P4b, and P4
The point c approaches infinitely close to each other, and it is not necessary to take the image twice. In such a case, the second side camera 22 and the third side camera 23 may simultaneously capture the images at the intermediate point between the points P4b and P4c by a determination step (not shown).

When the L2a size shown in FIG. 9 is transmitted from the microprocessor 40c to the microprocessor 40d, the required moving distance L3 can be calculated on the microprocessor 40d side. L3 may be transmitted to the microprocessor 40d. Note that the comparison determination in step 516 is
This is performed twice in total corresponding to the second side camera 22 and the third side camera 23, but when the molded product 30 is viewed from four directions, no matter which direction the abnormality occurs, the third abnormality detection output ( Step 517) will operate.

Furthermore, when the photographing point error ΔX is transmitted to the microprocessors 40d to 40c in step 505,
In step 404 of FIG. 10, the correction coefficient K is corrected,
An appropriate correction coefficient K is calculated by repeating this feedback operation in the initial stage of the learning operation.

Third Embodiment A visual inspection system for a high-speed moving body according to Embodiment 3 of the present invention will be described with reference to the drawings. FIG. 12 shows the third embodiment of the present invention.
It is a figure which shows the front structure of the external appearance inspection system of the high-speed moving body which concerns on.

In FIG. 12, 1a, 2a, 3a, ...
・ Is a multi-section molding machine that uses molds of the same shape,
Reference numeral 10 is a conveyor that is driven in the direction of arrow 13 by a motor 11, and 12 is an encoder that measures the amount of movement of the conveyor 10.

Molded products 1c, 1d, 1e, 2c, 2 molded by the above molding machines 1a, 2a, 3a, ...
d, 2e, 3c, 3d, 3e, ... Are mounted on the conveyor 10 by the mounting mechanism described below, and are moved as a large number of moving bodies (molded products) 30 in the directions indicated by arrows 13 in FIGS. Has become.

Each of the molding machines 1a, 2a, 3a, ... According to the third embodiment has three dies and performs three-piece molding. The molded products 1c, 1d, and 1e immediately after processing are processed by the arrow 1 by the mounting mechanism configured by the three partition plates 14, the connecting plate 15 connecting the partition plates 14 and the extruding rod 16.
It is transferred in the directions b, 1f and mounted on the conveyor 10.

The mounting mechanism provided in each of the molding machines 1a, 2a, 3a, ... Is controlled so that the molded products do not overlap on the conveyor 10. When the mounting operation is performed, the closing detection switch 1g is operated. , 2g, ... Are operated.

FIG. 13 shows the third embodiment of the present invention.
It is a figure which shows the structure of the external appearance inspection apparatus of the external appearance inspection system of the high speed moving body which concerns on FIG.

The differences from FIG. 2 will be mainly described. FIG.
, 200C is a first inspection unit of the appearance inspection device 32C. The first inspection unit 200C includes a microprocessor (CPU) 40e and a system memory (ROM) 41.
e, an arithmetic memory (RAM) 42e as a main component, and a closing detection switch 1g, 2 as a mounting signal.
g, ... Are connected.

Further, in the figure, reference numeral 300C denotes a second inspection section of the appearance inspection apparatus 32C.
C is a microprocessor (CPU: abnormality detection means) 4
0f, a system memory (ROM) 41f, and an arithmetic memory (RAM) 42f.

The microprocessor 40e executes the operation shown in FIG. 6 and the microprocessor 40f executes the operation shown in FIG. 7, but in the third embodiment described here,
In addition to the operation shown in FIG. 6, the microprocessor 40e is configured to execute the operation shown in FIG. 14 by time division operation.

Next, the operation of the appearance inspection system for a high-speed moving body according to the third embodiment will be described with reference to the drawings.

FIG. 14 is a flowchart showing the operation of the microprocessor of the first inspection section of the appearance inspection apparatus of the appearance inspection system for high-speed moving objects according to the third embodiment of the present invention.

In step 100 of FIG. 14, the microprocessor 40e starts its operation.

Next, in step 101, it is determined whether any of the closing detection switches 1g to 9g has operated.

Next, at step 102, if YES at step 101, the closing detection switch 1
The mold numbers of the molded products input from the section numbers assigned to g to 9g are stored in a shift register (not shown). For example, in section number 1g, since three dies are used, the die numbers are stored as 11, 12, and 13. Similarly, in section number 2g, since three molds are used, mold numbers are 21, 22, and 23.
, And the upper digit of the die number indicates the section number.

Next, in step 103, when the result in step 101 is NO, or in step 1 above.
Subsequent to 02, it is determined whether or not the movement signal of the conveyor 10 by the encoder 12 has been received.

Next, in step 104, if YES in step 103, the above step 102
Step the storage memory address of the mold number stored in.

Next, in step 105, when the result in step 103 is NO, or in step 10 above.
After step 4, it is determined in steps 215 and 219 of FIG. 6 and step 317 of FIG. 7 whether an abnormality detection output is generated.

Next, in step 106, if YES in step 105, the mold number information at the address position corresponding to the position of the abnormally formed molded product on the conveyor 10 is displayed on the abnormal alarm indicator 33. Read and display (alarm display means). In this step 106, depending on whether the cause of the abnormality is step 215 or step 219 in FIG. 6 or step 317 in FIG. 7, an abnormal segment display output or an alarm buzzer output is generated. .

Next, in step 107, if the answer in step 105 is NO, or following step 106, the conveyor 10 of the molded product in which an abnormality has occurred is generated.
It is determined whether or not the upper position has reached the discharge position where the discard nozzle 31 is installed.

Next, at step 108, if YES at step 107, the discard nozzle 31 is driven (discard nozzle driving means).

Then, in step 109, when the result of step 107 is NO, or following step 108, the operation is ended. When the operation is completed, the operation start step 100 is performed again.

The shift register for storing the die number stores the position of the molded product (moving body) 30 from the most upstream position of the conveyor 10 to the position where the waste nozzle 31 is installed, and the downstream thereafter. The position information need not be stored.

The steps 102, 103 and 104 constitute the tracking means 110.

Other Embodiments. The camera described in each of the above embodiments is assumed to be a CCD camera that detects infrared rays generated by a scorched molded product, and in the case of a molded product that does not emit light by itself, a projector (not shown) is also used. Is.

Further, in order to take a close-up photograph of a scorched molded product, as shown in FIG. 15, a surveillance camera 51 such as a top camera or a side camera is housed in an air-cooling box 50,
If the object is monitored through the heat-resistant transparent glass 52, the screen resolution can be increased. 5
3 is an air supply port connected to a duct not shown, 54 is an air supply direction arrow, 55 is an exhaust port connected to a duct not shown, and 56 is an exhaust direction arrow.

The side-face contour image comparison means shown in step 316 of FIG. 7 or step 516 of FIG. 11 assumes comparison by a known pattern matching method, and learning memory data to be compared or inspection target data. The image data and the like are the side contour image itself of the moving body (molded product). However, it is also possible to perform storage and comparison using typical digitized data such as maximum width, minimum width, height of the contour image, and deviation of the center line.

In step 215 of FIG. 6 and step 415 of FIG. 10, when the center point position of the upper surface contour image of the molded product 30 and the center line position of the conveyor 10 are largely deviated, it is determined that the molded product 30 is in abnormal position. (1) An abnormality detection output is generated.In addition to this, if the interval between multiple molded products is abnormally close or falls, it is further classified as an abnormality of the mounting mechanism. It is possible to add a function for detecting the state by using. Further, although the center point position of the top surface contour image is used to determine the deviation from the center line of the conveyor 10, the deviation may be determined based on the estimated bottom surface center position obtained by multiplying this by the correction coefficient.

In addition, the learning switch operation is performed by the learning switch 34.
The learning operation is completed automatically when a predetermined amount of production after the start of operation or when a predetermined number of products have been produced, or when the learning memory information becomes sufficient, the learning operation is completed. Then, the normal operation may be automatically started.

[0209]

According to the appearance inspection system for a high-speed moving body according to claims 1 to 3 of the present invention, as described above, the moving body conveyed at a high speed by the conveyor has the first camera for monitoring the upper surface and the side surface. It is monitored by a second camera for monitoring, the first camera monitors the top shape of the moving object for abnormalities, and the position information on the conveyor is obtained to determine the shooting timing of the second camera. The side surface shape is monitored for abnormality, and when an abnormality occurs, an upstream countermeasure output or a downstream countermeasure output is generated. Therefore, even for a moving body whose position on the conveyor is uncertain, it is possible to detect abnormalities in the shape and position of the moving body at high speed and with high accuracy, and promptly implement upstream and downstream countermeasures to reject defective products. This has the effect of reducing the number of products manufactured.

According to the appearance inspection system for a high-speed moving body according to claim 4 of the present invention, as described above, a plurality of second side monitoring cameras are used. Therefore, there is an effect that the abnormality of the side surface shape can be promptly and accurately determined by sharing and monitoring the side surfaces of the moving body in multiple directions.

According to the appearance inspection system for a high-speed moving body according to the fifth aspect of the present invention, as described above, the second camera for side surface monitoring captures the moving body at a plurality of timings. Therefore, by monitoring the side faces of the moving body in multiple directions with a small number of cameras, it is possible to inexpensively and accurately determine the side face abnormality.

According to the appearance inspection system for a high-speed moving body according to claim 6 of the present invention, as described above, the installation positions of the first camera and the second camera are separated, and the side image is photographed in the top image. It will be carried out after shooting. Therefore, after the position information of the moving body is obtained by the first camera, there is a time margin for calculating and calculating the timing of the side face shooting by the second camera. Therefore, there is an effect that the abnormality determination can be performed at high speed and with high accuracy.

According to the appearance inspection system for a high-speed moving object according to claim 7 of the present invention, as described above, the first camera for monitoring the upper surface of the high-speed moving object can move the moving object at a predetermined time or every predetermined moving amount of the conveyor. It is configured to photograph the top surface. Therefore, for a moving body at an unspecified position on the conveyor, even if no other optical sensor for shooting position detection is provided,
There is an effect that the omission of the upper surface photographing by the first camera does not occur.

According to the appearance inspection system for a high-speed moving body in accordance with claim 8 of the present invention, as described above, the second side-viewing camera is used from the top surface photographing of the moving body by the first camera to the side-face photographing position. The side surface of the moving body is photographed in the vicinity of the reference point at which the required moving distance is measured. Therefore, there is an effect that all the side surface contour images are converted into reference point positions and stored, and the judged contour image information and the reference contour image information are unified and handled to enable high-speed and highly accurate abnormality judgment.

According to the appearance inspection system for a high-speed moving body according to claim 9 of the present invention, as described above, the abnormality detection of the upper surface shape by the first camera is performed by comparing the learned memory information with the actual driving information. . Moreover, the learning information is replaced and stored with a large number of fixed point coordinate position information,
This stored information is reconverted into the center point coordinate position of the moving body actually measured in the practical operation, and is used as the comparison reference information. Therefore, even if the upper surface contour image changes at an unspecified position on the conveyor, there is an effect that the upper surface shape can be accurately determined using the learning information of multiple points.

According to the appearance inspection system for a high-speed moving body according to claim 10 of the present invention, as described above, the abnormality detection of the upper surface shape by the first camera is performed by comparing the learning memory information with the actual driving information. Be seen. Moreover, as the first camera for upper surface monitoring, a plurality of cameras installed at mutually different positions are used, and the upper surface contour image information at the adjacent fixed point coordinates is immediately calculated by interpolation calculation from the plural upper surface contour images by the plurality of cameras. The average of many calculated information is learned and stored. Further, even in the practical operation stage, since the upper surface contour image information at the adjacent fixed point coordinates can be immediately obtained from the plural upper surface contour images by the plural cameras by the interpolation calculation, the measurement conversion information and the learning memory information are compared to detect the abnormality of the upper surface shape. It is configured so that it can be judged. Therefore, even if the upper surface contour image changes at an unspecified position on the conveyor, there is an effect that the upper surface shape can be promptly determined using the learning information of multiple points.

According to the appearance inspection system for a high-speed moving body according to claim 11 of the present invention, as described above, the abnormality detection of the side surface shape by the plurality of second cameras is performed by comparing the learning memory information with the actual driving information. Done by Moreover,
The side image of the moving body is taken by the second camera in the vicinity of the reference point based on the position information of the moving body by the first camera, and the side surface contour image is centrally converted to this reference point position for handling. There is. Therefore, there is an effect that the abnormality of the multi-directional side surface shape of the moving body can be determined at high speed and with high accuracy.

According to the appearance inspection system for a high-speed moving body according to claim 12 of the present invention, as described above, the abnormality of the side surface shape is detected by the second camera by comparing the learning memory information with the actual driving information. Be seen. Moreover, the second camera is adapted to perform side view photography twice at different positions for one moving body. Therefore, it is possible to inexpensively and highly accurately determine the abnormality of the multidirectional side surface shape of the moving body by using a small number of second cameras.

According to the appearance inspection system for a high-speed moving body according to claim 13 of the present invention, as described above, when the abnormality of the side surface shape is detected by the second camera, the plurality of side surface monitoring second cameras are provided. It is configured to use side-view contour image information of different viewpoints, which are actually measured and photographed, and convert the side-view contour image information on the reference point to each other for comparison. Therefore, there is an effect that the abnormality of the side surface shape of the moving body can be determined at high speed and with high accuracy without the need for learning and memory.

According to the visual inspection system for a high-speed moving body in accordance with a fourteenth aspect of the present invention, as described above, the high-speed moving body conveyed by the conveyor is a molded product formed by a molding machine having a plurality of sections. Then, tracking means for moving the section number or mold number of the applied molding machine is added. Therefore, the section number or mold number of the molding machine that molded the abnormally detected molded product is read without the need to read the mold number, etc. provided on the bottom or side of the molded product that is difficult to detect in the upstream stage. There is an effect that can be displayed.

As described above, according to the visual inspection system for a high-speed moving body according to the fifteenth aspect of the present invention, when the abnormal position of the moving body on the conveyor or the abnormal shape of the upper surface or side surface is detected, the moving body is detected. The discharge processing, molding machine section number or mold number, and abnormal classification information are read out and displayed. Therefore, in addition to the discharge processing of the moving body itself in which an abnormality has occurred, it is possible to identify which mold is abnormal or which section mounting mechanism is abnormal, and promptly take upstream measures to eliminate defective products. There is an effect that can be reduced.

According to the appearance inspection system for a high-speed moving body according to claim 16 of the present invention, as described above, the first and second cameras for monitoring the moving body are housed in the air-cooled box and the heat-resistant transparent glass is used. It is configured to take a photograph of a molded product that is a moving body via. Therefore, there is an effect that it is possible to obtain a clear image by close-up photography with respect to a high-temperature moving body and perform highly accurate abnormality determination.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a visual inspection system for a high-speed moving body according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of an appearance inspection device of an appearance inspection system for a high-speed moving body according to Embodiment 1 of the present invention.

FIG. 3 is a diagram showing a field plane of a top camera of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

FIG. 4 is a diagram showing fixed point coordinates and interpolation straight lines of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

FIG. 5 is a side view showing the arrangement of side cameras of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

FIG. 6 is a flowchart showing the operation of the microprocessor of the first inspection unit of the appearance inspection device of the appearance inspection system for a high-speed moving body according to Embodiment 1 of the present invention.

FIG. 7 is a flowchart showing an operation of the microprocessor of the second inspection unit of the appearance inspection device of the appearance inspection system for a high-speed moving body according to the first embodiment of the present invention.

FIG. 8 is a diagram showing a configuration of an appearance inspection device of an appearance inspection system for a high-speed moving body according to Embodiment 2 of the present invention.

FIG. 9 is a diagram showing a field plane of a top camera of the appearance inspection system for a high-speed moving body according to Embodiment 2 of the present invention.

FIG. 10 is a flowchart showing the operation of the microprocessor of the first inspection unit of the appearance inspection device of the appearance inspection system for a high-speed moving body according to Embodiment 2 of the present invention.

FIG. 11 is a flowchart showing the operation of the microprocessor of the second inspection section of the appearance inspection device of the appearance inspection system for a high-speed moving body according to Embodiment 2 of the present invention.

FIG. 12 is a diagram showing a configuration of a visual inspection system for a high-speed moving body according to a third embodiment of the present invention.

FIG. 13 is a diagram showing a configuration of an appearance inspection device of an appearance inspection system for a high-speed moving body according to Embodiment 3 of the present invention.

FIG. 14 is a flowchart showing the operation of the microprocessor of the first inspection unit of the appearance inspection device of the appearance inspection system for a high-speed moving body according to Embodiment 3 of the present invention.

FIG. 15 is a diagram showing a camera mechanism of a visual inspection system for a high-speed moving body according to another embodiment of the present invention.

[Explanation of symbols]

1a, 2a ... 9a Molding machine, 1c, 1d, 1e Molded product (moving body), 1g, 2g ... 9g Input detection switch (mounting signal), 10 conveyor, 12 encoder (moving signal), 20 Top camera (First camera), 21
First side camera (second camera), 22 Second side camera (second camera), 23 Third side camera (second camera), 24 Fourth side camera (second camera), 25 Second top camera (second camera) 1 camera), 30 molded products (moving body), 31 waste nozzles (downstream countermeasure output means), 32
A, 32B, 32C Visual inspection device, 33 Abnormality alarm indicator (upstream countermeasure output means), 40a CPU (abnormality detection means), 40b CPU (abnormality detection means), 40c C
PU (abnormality detecting means), 40d CPU (abnormality detecting means), 40e CPU (abnormality detecting means, tracking means), 40f CPU (abnormality detecting means).

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ identification code FI theme code (reference) G06T 7/60 300 G06T 7/60 300A 5L096 (72) Inventor Noriko Nagata 2-2 Marunouchi, Chiyoda-ku, Tokyo No. 3 Sanryo Electric Co., Ltd. F term (reference) 2F065 AA02 AA51 BB06 BB15 DD06 FF04 JJ03 JJ05 JJ07 JJ26 MM02 PP15 QQ24 QQ25 QQ31 QQ33 RR03 SS09 TT03 2G051 AA12 AB23 CA41 CA22 CA23 CC03 DA11 EA12 EA19 4G015 GA01 5B057 AA02 AA19 BA02 DA03 DA07 DA15 DB02 DC03 DC06 DC16 DC33 DC40 5L096 BA03 CA05 DA03 FA06 FA62 FA64 FA69 KA15

Claims (16)

[Claims] 1. A first camera for upper surface monitoring, which photographs a high-speed moving body conveyed by a conveyor and monitors the upper surface shape of the moving body and a passing position on the conveyor, and a first camera for conveying the conveyor. A second camera for side surface monitoring, which captures high-speed moving objects, obtains passing position information from the first camera, and monitors the side surface of the moving object for abnormality, and the first camera and the second camera. Abnormality detection means for detecting the shape and state abnormality of the moving body based on the captured image of the moving body, upstream countermeasure output means for the conveyor driven by the abnormality detection means, and driven by the abnormality detection means A visual inspection system for a high-speed moving object, comprising: downstream countermeasure output means for the conveyor. 2. A first camera for monitoring an upper surface, which photographs a high-speed moving body conveyed on a conveyor and monitors for an abnormality in the upper surface shape of the moving body and a passage position on the conveyor, and is conveyed on the conveyor. A second camera for side surface monitoring, which captures high-speed moving objects, obtains passing position information from the first camera, and monitors the side surface of the moving object for abnormality, and the first camera and the second camera. A high speed characterized by comprising an abnormality detecting means for detecting an abnormality in the shape and state of the moving body based on the captured image of the moving body, and an upstream countermeasure output means for the conveyor driven by the abnormality detecting means. Appearance inspection system for moving objects. 3. A first camera for upper surface monitoring, which photographs a high-speed moving body conveyed by a conveyor, and monitors abnormality of the upper surface shape of the moving body and a passage position on the conveyor, and is conveyed by the conveyor. A second camera for side surface monitoring, which captures high-speed moving objects, obtains passing position information from the first camera, and monitors the side surface of the moving object for abnormality, and the first camera and the second camera. A high speed characterized by comprising an abnormality detection means for detecting an abnormality in the shape and state of the moving body based on the captured image of the moving body, and a downstream countermeasure output means for the conveyor driven by the abnormality detecting means. Appearance inspection system for moving objects. 4. The second camera, wherein a plurality of cameras installed at different viewing angles are used to perform multi-directional side surface monitoring of the moving body. 3. A high-speed moving body appearance inspection system according to 3. 5. A multi-directional side surface monitoring of the moving body is performed by taking a side image of the moving body by the second camera at a plurality of timings when the relative positions of the second camera and the target moving body are different. The appearance inspection system for a high-speed moving body according to claim 1, 2 or 3, wherein 6. The first camera is installed at a position upstream of the conveyor as compared with the second camera, and after the first camera first captures an upper surface image of the same moving body,
The appearance inspection system for a high-speed moving object according to claim 1, 2, or 3, wherein a side image is captured by the second camera when the moving object moves to the position of the second camera. 7. The first camera photographs the top surface of the moving body for a predetermined time or for each predetermined movement amount of the conveyor,
7. The appearance inspection system for a high-speed moving body according to claim 6, wherein at least one top surface photographing is performed when the moving body is within the field of view of the first camera. 8. The side surface of the moving body of the second camera near the side surface reference point, which is the time point when the required moving distance from the top surface shooting time of the moving body to the side shooting position of the moving body is measured by the first camera. The image of the subject is taken.
Appearance inspection system for high-speed moving objects. 9. The anomaly detection means includes a primary learning storage means for storing a large number of pieces of upper surface contour image information that are learned and photographed by the first camera and have a center point of the upper surface contour image at various coordinate positions, and the primary learning storage. Secondary learning storage means for storing contour image information at various fixed point coordinates interpolated from the contour image information stored by the means, and stored contour image information for judgment at unspecified coordinate positions actually measured and photographed in the operation stage Primary measurement storage means, and learning read storage means for calculating and storing contour image information that should be at the unspecified coordinate position by interpolation calculation from contour image information at adjacent fixed point coordinates stored in the secondary learning storage means, If the comparison deviation between the comparison means for comparing the contour image information stored by the learning read storage means and the primary measurement storage means and the comparison means is excessive. Top shape abnormality top shape for generating an output alarm output means and the visual inspection system of claim 1, 2 or 3 high-speed moving body according to characterized in that it has a. 10. A plurality of cameras installed at mutually different positions are used as the first camera, and the abnormality detection unit is configured to learn and photograph by the first camera, and a center point of an upper surface contour image has various coordinate positions. Primary learning storage means for storing a plurality of upper surface contour image information, and secondary learning storage means for storing contour image information at adjacent fixed point coordinates interpolated from the contour image information stored in the primary learning storage means. A tertiary learning storage means for storing contour image information at adjacent fixed point coordinates averaged from the contour image information stored in the secondary learning storage means; Primary measurement storage means for storing determination contour image information, and a contour image at adjacent fixed point coordinates interpolated from the contour image information stored in the primary measurement storage means Secondary measurement storage means for storing image information; comparison means for comparing contour image information at adjacent fixed point coordinates stored by the tertiary learning storage means with contour image information stored by the secondary measurement storage means; 4. The appearance inspection system for a high-speed moving body according to claim 1, further comprising an upper surface shape abnormality output means for generating an upper surface shape abnormality output when the comparison deviation by the comparison means is excessive. 11. The anomaly detection means includes a primary learning storage means for storing side face contour image information learned and photographed by a plurality of second cameras, and a center point of the contour image stored by the primary learning storage means. Secondary learning storage means for storing the transformed contour image information when moved to the common coordinate line, and average contour image information calculated as an average value of a large number of transformed contour image information stored by the secondary learning storage means. A third-order learning storage means for storing the following, a primary measurement storage means for storing a side-face contour image actually measured and photographed in the operation stage, and a center point of the contour image stored by the primary measurement storage means on the predetermined common coordinate line. The secondary actual measurement storage means for storing the converted contour image information when moving, the contour image information stored by the tertiary learning storage means, and the secondary actual measurement storage means. And a side shape abnormality output means for generating a side shape abnormality output when the comparison deviation by the comparison means is excessive. The appearance inspection system for a high-speed moving body according to 2 or 3. 12. The abnormality detecting means defines a primary learning storage means for storing side face contour image information learned by the second camera, and a center point of the contour image stored by the primary learning storage means. Secondary learning storage means for storing the transformed contour image information when moved to the common coordinate line, and average contour image information calculated as an average value of a large number of transformed contour image information stored by the secondary learning storage means. Third-order learning storage means to be stored, first-side actual measurement storage means to store a side-face contour image actually measured and photographed in an operation stage, and a center point of the contour image stored by the first-order actual-measurement storage means is moved to the predetermined common coordinate line. The secondary measurement storage means for storing the converted contour image information at the time, the contour image information stored by the tertiary learning storage means, and the secondary measurement storage means. A comparison means for comparing stored contour image information, and a side shape abnormality output means for generating a side shape abnormality output when the comparison deviation by the comparison means is excessive, and the second camera has one The side surface photographing is performed twice at different positions with respect to the moving body, and the predetermined common coordinate line is set at two positions corresponding to the different positions. Visual inspection system for high-speed moving objects. 13. The abnormality detecting means includes a plurality of primary measurement storage means for storing side contour image information actually measured and photographed by a plurality of second cameras; and a side contour image stored by the plurality of primary measurement storage means. A plurality of secondary measurement storage means for storing the converted contour image information when the center point of is moved to a predetermined common coordinate line, and the contours of the plurality of second cameras stored by the plurality of secondary measurement storage means 4. The high-speed according to claim 1, 2 or 3, further comprising: comparing means for comparing image information; and side shape abnormality output means for generating a side shape abnormality output when the comparison deviation by the comparing means is excessive. Appearance inspection system for moving objects. 14. A molding machine for a plurality of sections using molds of the same shape, a conveyor for mixing and carrying molded products molded by the molding machine, and a molded product as a moving body mixed and carried by the conveyor. A first camera for upper surface monitoring, a second camera for side surface monitoring of a molded product as a moving body mixed and conveyed by the conveyor, and the image of the molded product photographed by the first camera and the second camera, An abnormality detecting means for detecting a shape abnormality of the molded product, a tracking means for moving the section number or mold number of the molding machine applied based on the mounting signal of the molded product and the movement signal of the conveyor, and the abnormality is detected. The section number or mold number of the molding machine that molded the molded product is read and displayed. An appearance inspection system for high-speed moving objects, which is equipped with the function to detect abnormalities in the molding die by monitoring the shape of the molded product immediately after molding. 15. The abnormality detecting means detects that a center position of an upper surface image of a molded product as a moving body photographed by the first camera deviates from a movement center line of the conveyor by a predetermined value or more. A first abnormality detection output means for generating a first abnormality detection output, a second abnormality detection output means for generating a second abnormality detection output related to the abnormality of the upper surface image of the molded product, and a side image of the molded product Third abnormality detection output means for generating a third abnormality detection output related to abnormality, section number or mold number information of the molding machine that molded the molded product in which the abnormality was detected, and the first, second and third abnormality classifications It has alarm display means for reading out and displaying information on the abnormality alarm display, and waste nozzle driving means for driving a waste nozzle for eliminating the abnormal molded product from the conveyor when each abnormality occurs. Fast mobile visual inspection system of claim 14 wherein wherein the symptoms. 16. The first camera or the second camera for monitoring the moving body is housed in an air-cooling box through which cooling air is supplied and exhausted, and a high temperature is provided through a heat-resistant transparent glass provided on a wall surface of the air-cooling box. 15. The appearance inspection system for a high-speed moving body according to claim 1, 2, 3 or 14, wherein a close-up image of a molded product, which is the moving body, is taken.