JPS60249559A - Vision tracking apparatus - Google Patents

Abstract

PURPOSE:To reduce the number of position recognizing sensors while to detect the abnormality in the tracking system by tracking the material on all production and machining lines then taking the vision of the material and correcting the moving position. CONSTITUTION:In a tracking system for monitoring the passing position of material in a production line having several kind of processing and machining processes while provided with material passing position recognizing sensors and operated with predetermined speed, the signal from a video recognizer 9 is selected through a distributor 12 and fed to an image processor 13. The signal applied with recognition processing through said processor 13 is operated through a processor 5 and fed to a video unit 14 to display the image recognized material onto a display 8. The signal from a line speed detector 11 is taken into the processor 5. The distributor 12 is controlled by the processor 5. In such a manner, the shape and the position of a plurality of moving objects having different shape on the production line can be recognized sequentially.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a device for tracking a moving object such as a material in a continuous process line such as manufacturing or processing, and in particular,
The present invention relates to a vision tracking device suitable for sequentially recognizing the shapes and positions of a plurality of moving objects having different shapes.

[Background of the invention]

Conventionally, in a tracking system, as shown in Figs. 1 and 2, a position recognition sensor 2 installed on a production line 1 detects a passing point of a moving object (hereinafter referred to as material) 3, and a detection signal is input to the sensor. .

A method has been adopted in which the position of the material 3 is evaluated or recognized using the device 4 and the processing device 5. According to this method, since the material 3 can only be detected at the position where the position recognition sensor 2 is installed, there is a drawback that the section where the position recognition sensor 2 is not installed is not monitored. Furthermore, since the position recognition sensor 2 is usually a point detector and can only detect the material 3 as a point, it has been impossible to recognize the shape of the material 3.

Therefore, this method cannot be applied to a mixed line in which materials 3 of different shapes are processed on the same production line 1.

In addition, in this method, when tracking the position of the material 3, the number of position recognition sensors 2 becomes very large, and the size of the census force device 4 increases accordingly.Furthermore, the position of the material 3 In addition, it was possible to perform pattern matching by recognizing the shape of the material 3 and correct the position of the material 3 at the predicted time.

[Purpose of the invention]

An object of the present invention is to track materials in a manufacturing or processing line throughout the line.

It is an object of the present invention to provide a vision tracking device that can recognize the shape of a material, reduce the number of position recognition sensors, and further correct the moving position of the material or detect abnormalities.

[Summary of the Invention] In the production line 1 shown in FIG. 1, tracking of the material 3 moving on the production line 1 is essential. The position of the material 3 can be recognized by the position recognition sensor 2. Conventionally, the position recognition sensor 2 is a material position indicator attached to the simulated production line on the display device 8, as shown in FIG. A method was adopted in which the position of material 3 was confirmed by blinking 7. According to this method, the position of the material 3 between the material position indicators 7 cannot be monitored at all. On the other hand, when performing such tracking,
Normally, the speed of the production line 1 is known in a certain processing section, for example the section X in FIG. 3, and can be a source of information. That is, at a certain time T. In this case, the position of the material 3 introduced at point A can be accurately determined based on the line speed. Therefore, by installing a visual recognizer at point A, its shape can be recognized, specific fixed points of material 3 (points whose shape does not change due to processing, etc.) can be detected, and its position can be predicted in time series. It becomes possible to do so.

The material 3 whose position has been accurately predicted in this way is displayed as a moving image on the display [18] by the video device, and moves on the simulated production line. With this method, materials on the simulated production line can be checked without interruption.

In addition, in a mixed line where materials 3 of different shapes are processed on the same production line 1, by changing the above-mentioned invariant points, they can be recognized separately, and different symbols can be recognized on the simulated production line. can be distinguished using.

On the other hand, 12 visual recognizers are installed in each processing section on the production line.
As a result, the number of points is only a few to a few points at most, which is extremely reduced compared to the point detector described above.

Further, assuming that the material 3 introduced at point A at time To reaches point B at time T1 predicted from the line speed and processing section length, the visual recognizer at point B An image like the one shown is obtained. In Figure 4,
It becomes possible to correct the image position (predicted position) stored in the processing device based on the deviation ΔX between the feature points of the actually recognized image (solid line) and the predicted image (broken line). In this case, by providing a regulation for Ax, it is possible to further detect abnormalities in the system.

[Embodiments of the invention]

An embodiment of the present invention will be described below.

On the manufacturing line 1 having different processing sections X1 to X11 as shown in FIG. 5, the material 3 moves from point A through each processing section to point Z as shown by the broken line. The visual recognizer 9 is installed at the beginning of the processing sections X1 to Xll, and recognizes the shape and position of the material 3.

That is, a material 3 having feature points 10 as shown in FIG.
can be captured by the visual recognizer 9 as a plane image as shown in FIG. Visual recognizer p1. Since the coordinates of p on the production line 1 are usually known, if its relative position (distance) is now X, the material 3 recognized at P at time T0 is at the line speed of processing section X1, That is, due to the velocity V of the material, it enters the field of view of PI3 at time T.

Time Tl = T g + x 1 / v t (1)
Therefore, by recognizing the image of P2 at time T, it is possible to correct the predicted position of material 3. Furthermore, when the processing section X is divided into n1 pieces, the unit section distance ΔX on Xl is the time ΔT for the material 3 to move by Δx1 from equation (2).

teeth, ΔT, = Δx, / vs (3) And the position lxα of material 3 at time T, α is.

T1 α=α・ΔT,=α・ΔXt/Vt (4)Xα
=v, XT, α (5). Therefore, by incrementing α, the position of the material at any given time can be known.

Next, based on this idea, the tracking device for material 3 will be explained.

FIG. 18 is a configuration diagram of the tracking device. The signal from the visual recognizer 9 is selected by a distributor 12 and sent to an image processing device 13. The signal that has been recognized and processed by the image processing device 13 is calculated by the processing device 5 and sent to the video device 14.

The video device 14 displays one image of the recognized material 3 on the display device 8. The signal of the line speed detector 11 is processed by the processing device 5
be taken in. Further, the 1-distributor 12 is controlled by the processing device 5.

FIG. 9 shows an image on the display device 8. The simulated production line 6 can display the simulated material 15 at any position.

As the material 3 moves on the production line 1, the display position of the simulated material 15 on the simulated production line 6 changes to A-+B-+
It changes sequentially from c...→L.

FIG. 10 shows a signal processing block diagram of the visual recognizer 9.

The signal from the visual recognizer is A/D converted and subjected to region segmentation processing. FIG. 11 shows the area division method. The A/D converted signal is binarized using a predetermined threshold. The binarized signal is divided into areas ■, ■, ■ as shown in Fig. 11.
divided into

After the region is divided, region ■ is extracted. This is done by knowing the area and length of the area (3) in advance through measurement or the like, and extracting the feature amounts as the area and circumference.

The center of gravity of the extracted region (■) can be determined.

On the other hand, since the center of gravity of the area (2) can be similarly determined from the area and circumference, the center of gravity of the area (2) can now be set to the origin O at the starting point A of the manufacturing line 1. Therefore, as shown in FIG. 11, the center of gravity of the area (2) is expressed as a relative value to the origin O (xo.

It becomes possible to display the coordinates of yo).

After the coordinates of the material 3 are determined in this manner, the position of the material 3 is predicted using the velocity signal. The material 3 recognized at time T0 changes the production line speed in processing section A-B to v (t
), it will advance only by 77 intervals. Therefore, the simulated material 15 is placed at time T. After projecting on the simulated production line 6 at time T, as shown in FIG. After +Δτ, it becomes possible to project the image at the relative position ΔX with time T. X
By setting T small, the simulated material 15 becomes a moving image that yields continuously. This situation is shown in FIG. The creation of this image is sequentially performed in process 5.

Normally, the production line 1 moves in only one direction, but the visual recognizer 9 detects the y of the feature point 10 as shown in FIG.
Since the coordinates in the direction can also be known, the coordinates in the y direction can also be created in the same way as creating the coordinates in the x direction. That is, the movement amount Δy in the y direction is calculated by equation (7).

However, Vγ(1) is the speed in the y direction.In this way, the simulated material 15 moves on the simulated production line 6 displayed on the display device 8 by the processing device 5 and the video device [14, but now at time T1 Assume that material 3 reaches point B. The time T1 can start from the distance X1 between A and B. At time T, the processing device 5 predicts that the material 3 has reached point B, and the visual recognition device 9 at point B can capture its image.

FIG. 14 shows an image at time T1. At time T, the feature point 1o of the material 3 is (Xl.

y,).

On the other hand, the predicted position of material 3 is (x%).

yI'), so we can calculate the deviation and ε here.

ε = x, -X1 (8) 6 = A, -7, (9) The processing device 5 evaluates using the deviations ε and ε determined in this way. That is, IEx l, Iε a
1 is the tolerance range ε. Eh, ε. 7 or less, the position of the simulated material 15 is corrected and the actually captured image (Xs
*'ix). On the other hand, 1ε, 1,1εy 1 is ε.

8. If it is ε, A or more, it is determined that the production line 1 is abnormal. In this way, the images obtained by each visual recognizer 9 can be evaluated.

Next, consider a case where the deviation ε is large and no image is obtained due to P at time T1. In this case, the material 3 has not reached point B or has passed through point B at time T1. The position at this time can be determined by confirming the value 15 using equation (6), v(x,
) abnormalities can be determined.

When the manufacturing line 1 shown in FIG. 5 is a continuous process line without dead time, the visual recognizer 9 recognizes the processing sections Xi, X2, . ..., it is not necessary to install it for each X11. That is, at point A at time T. Since the material 3 recognized at point Z is accurately predicted and tracked by the method described above, it is only necessary to finally recognize it at point Z. Therefore, since only two visual recognizers, point A and point Z, are required, the number of visual recognizers can be extremely reduced.

Furthermore, consider a case where the manufacturing line 1 is a mixed line and processes multiple types of materials 3 having different shapes. In this case, it is necessary to consider feature points 1 and 0 that can be distinguished from each other. An example is shown in FIG. Feature points 10
Since the shape of can be determined arbitrarily, its feature quantity is also free.

By providing such feature points 10, the aforementioned predictive tracking is possible even if the manufacturing line 1 is a mixed line, and the simulated materials 15 on the simulated manufacturing line 6 can be displayed using symbols that can be distinguished from each other. That's a good thing.

〔Effect of the invention〕

According to the modified invention, since consumption can be captured as an image, continuous image tracking of the material, reduction in the number of material recognition sensors, prediction of the position of the material, and detection of abnormalities in the tracking system become possible.

[Brief explanation of drawings]

Figure 1 shows a conventional tracking manufacturing line;
3 is a diagram of a conventional tracking device, FIG. 4 is a diagram showing an image of a material obtained by a visual recognition device, and FIG. 5 is a diagram showing a tracking production line according to an embodiment of the present invention. FIG. 6 is a diagram showing an example of the material, FIG. 7 is an explanatory diagram of the characteristic points of the material, FIG. 8 is a configuration diagram of a tracking device according to an embodiment of the present invention, and FIG. 9 is a diagram showing an example of the present invention. A diagram showing a tracking display device, FIG. 10 is a processing block diagram of a visual recognizer signal, FIG. 11 is an explanatory diagram of region division of a material, FIG. 12 is a diagram showing temporal changes in feature points, and FIG. 13 is a diagram of a processing block diagram of a visual recognizer signal. Tracking image diagram, Figure 14 is an explanatory diagram of material position correction, Figure 1
FIG. 5 is an explanatory diagram of materials having mutually different feature points. 1... Production line, 3... Material, 9... Visual recognizer, 10... Feature point. Agent Patent Attorney Akio Takahashi $ + mouth 3 eyes ヰ 4 figures ki 50 - ko 2 well 3 eyes year 9 prison rod 10 mouth 3 bud n E] 7□ rod 13 eupushi 15 eyes

Claims (1)

[Scope of Claims] (1) A tracking device that monitors the passing position of the material on a production line that has several types of processing or processing steps, is equipped with a sensor for recognizing the passing position of the material, and is operated at a predetermined speed, a visual recognizer, an image processing device for processing recognition signals thereof, a video device and a display device for displaying a simulated image of the manufacturing line, a processing device for comprehensively controlling these, and an image recognition device. means for recognizing the shape and position of the material based on a recognition signal from the container, and sequentially changing the shape and position of the material on the production line displayed in a simulated manner on the display device connected to the video device; A vision tracking device characterized by: