JPH0355669A - Object recognizing device - Google Patents

Abstract

PURPOSE:To initialize a monitor area with high accuracy satisfactory for the practical use by setting a key initial setting range for each automatically set section and then setting a monitor area at a monitor point included in the corresponding initial setting range. CONSTITUTION:The video signal VD obtained from a TV camera 16 is supplied to a monitor 32 via a synthesizing circuit 31. Thus an image corresponding to a viewfield FILD is displayed on a display screen PIC as shown on a diagram. At the same time, a CPU 17 supplies the monitor area display data KDIS showing the monitor areas K1, K2... to a graphic display control part 33 and also applies a monitor area display signal VDK to the circuit 31. Thus the areas K1, K2... are superimposed on the image of the FILD on the screen PIC of the monitor 32. Then the CPU 17 produces the lightness information on the areas K1, K2... via a monitor data forming part 35 based on the video data DATA fetched by a recognizing information memory 19.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an object recognition device, and particularly to an object recognition device for inspecting the presence or absence of an object to be recognized, the presence or absence of defects, etc. based on a video signal that can be obtained from a video camera. It is suitable for application in various cases. [Summary of the Invention] The present invention provides an object recognition device using a television camera, in which a brightness that makes it easy to determine when an image of a recognition target is captured based on video information obtained from an initial setting range of vJ numbers that constitute an automatically set section is provided. By determining monitoring points with information and automatically initializing the monitoring area at the monitoring points,
The monitoring area can be initially set with a sufficiently high precision for practical use without placing an excessive burden on the operator.

[Conventional technology]

Conventionally, this type of object recognition device has been used to monitor the occurrence of abnormal operations during the injection molding cycle of injection molding machines, or to inspect products in the manufacturing process of printed wiring boards, etc. In the case of an injection molding machine, an image of the mold and its surroundings, and in the case of a printed wiring board, images on the printed wiring board are captured by a television camera to obtain a video signal representing the display screen PfC as shown in Fig. 17. A method has been proposed in which objects are recognized based on whether or not a predetermined change has occurred in the video signal (Japanese Patent Application Laid-Open No. 39581/1983).

In this case, a predetermined position PI, P2 within the display screen PIC represented by a video signal obtained from a television camera
Monitoring area KL K2 of a predetermined size...
..., and set these monitoring areas Kl, K2...
Obtain monitoring information regarding the brightness of the monitoring areas Kl, K2, etc. at each predetermined processing cycle by integrating the video signal corresponding to , and an unexpected change has occurred in the monitoring information at each processing cycle. It is determined that an abnormality has occurred in the recognition target. [Problem to be solved by the invention] However, when trying to recognize the presence or absence of an object or the presence or absence of a defect using such an object recognition device, the display screen P
Coordinate positions Pi (X+.)'+), P2 (x., y
2) If the settings for... are not appropriate, there is a risk of erroneous judgments in the recognition results. For example, when inspecting a printed wiring board, as shown in FIG. 18, the television camera 2 is placed so as to directly face the surface of the printed wiring board l to be recognized, and placed at a predetermined position determined by its field of view FILD. Positioned printed wiring board 1
The video signal VD is obtained by capturing an image of the video signal VD.
In such a case, when trying to determine whether or not a genuine component is correctly mounted on the printed wiring board 1, in order to reduce the error rate of the determination result as much as possible, it is necessary to It is considered that the monitoring regions K1, K2, . . . may be set in the image portions having brightness information that are easy to determine, that is, the brightest image portions or the darkest image portions.

By setting the monitoring areas Kl, K2, etc. in the brightest or darkest image part, if an abnormality such as missing parts occurs, the monitoring area will be removed. Since the brightness of areas K1, K2... changes from the brightest or darkest brightness to the brightness of the insulating substrate, which usually has a medium brightness, it is easy to judge the difference in brightness. On the other hand, if the monitoring areas K1, K2, etc. are set in the video part with medium brightness,
Even if a component is missing, the brightness only changes from medium to medium, so there is a high probability that an error will be made in determining an abnormality.

In addition, in an object recognition device applied to an injection molding machine monitoring device, as shown in FIG.
A television camera 11 is mounted on a mounting base 10 fixed on the holder 9 of the fixed side mold 8 in order to be able to capture an image of the movable side mold 5 and its surroundings within the field of view FILD when the mold is moved to the open state. When mounted, images of the movable side mold 5 and its surroundings are taken from diagonally above. In this case, when setting the monitoring areas Kl, K2, etc., the image part with the largest change in brightness in the relationship between the image before projection and the image after projection is By setting the monitoring areas K1, K2, . . . in this video portion as a “video portion having information on the image data,” the error rate of determination can be reduced.

Incidentally, when monitoring whether or not the injection molded product is correctly ejected from the mold during the ejection process, monitor nozzle areas K1, K2, If you set ..., and the image after projection changes to become darker (or brighter),
The probability of correctly determining the change in question increases. On the other hand, if the monitoring areas Kl, K2, etc. are set in a video part that has approximately the same brightness as the brightness of the image after projection, then the monitoring areas K1, K2, before and after projection are set.
Because the difference in brightness between ... is small, the probability of misjudgment increases.

As described above, in this type of object recognition device, whether or not the monitoring regions Kl, K2, etc. can be set at as appropriate positions as possible is one of the conditions that has a huge impact on the error rate of the judgment results. Conventionally, before starting the monitoring operation, the operator displays an image of the standard product to be monitored (this is called the master) on a monitor, and then selects the optimal product while watching the image on the monitor. Possible video positions are selected and monitoring areas Kl, K2, . . . are manually set.

However, when attempting to manually set the monitoring area in this manner, the setting work inevitably becomes extremely complicated. For example, when a large number of parts are mounted on a printed wiring board, setting the monitoring area becomes an excessive burden on the operator.

In addition, in the case of an ejection control aircraft monitoring device, when the brightness of the ejection control product is at an intermediate level, it is often difficult to manually select the image capture position with the lowest judgment error rate among them.

The present invention has been made with the above points in mind, and we would like to propose an object recognition device that can automatically initialize the monitoring area to a position where it is possible to obtain a sufficiently high judgment accuracy for practical use. That is.

[Means for solving problems]

In order to solve this problem, in the first invention, monitoring areas Kl, K2, . In an object recognition device that determines whether or not a recognition target is normal by setting ... and monitoring changes in brightness information DWCH obtained from the monitoring nod areas Kl, K2, ... With the master displayed on the display screen PIC, multiple initial setting ranges SET”i? configured automatic setting sections SECII to SEC?4, SEC4 are displayed on the display screen PIC.
1 and SEC4 2, and the master is displayed on the display screen PIC, the automatic setting sections SECII to SE
Find the brightness information DWCH for each of the initial setting ranges SET included in C34, SEC4 1, and SEC4 2, and among the obtained brightness information DWCH, select the initial setting range SET that has brightness information that makes it easy to determine whether it is normal or not. is determined as the representative initial setting range SET■, and monitoring areas K1, K2, etc. are set in the representative initial setting range SET■. Further, in the second invention, in addition to the first invention, monitoring areas Kl, K2, etc. are sequentially set with sample points included in the representative initial setting range SET and corners as monitoring positions, respectively. The second brightness information of each sample point is determined, and sample points having brightness information that makes it easy to determine whether or not the second brightness information is normal are selected from the monitoring areas K1, K2, etc.
The default position of ... is determined. [Effect]? In the invention of 1, the representative initial setting range SET■ is each automatic setting section fisEc11 to SEC34, SEC4 1
It has brightness information that is the easiest to judge among the SEC42 and SEC42, and therefore can reduce the error rate of judgment results, thereby reducing the burden on the operator in initial setting of the monitoring areas K1, K2, etc. It can be done. In addition, in the second invention, by being able to set the monitoring areas K1, K2, etc. directly to the monitoring area that has the brightness information that is easiest to judge among the representative initial setting range SET gates, errors in the judgment results can be avoided. The rate can be further reduced. [Embodiment] An embodiment of the present invention will be described in detail below with reference to the drawings. [1] Overall structure of the first embodiment In FIG. 1, 14 indicates an object recognition device as a whole, which is applied to monitor the ejection command machine body 15 described above with reference to FIG. 19. Television camera 16 has field of view FI
A video signal VD obtained by photographing and copying the movable side mold in the LD and its surroundings is output, and this video signal VD is converted into video data DATA in the video data input circuit 18 under the control of the CPU 17 and is stored in the recognition information memory. Let 19 take it in.

The CPU 17 executes data processing of object recognition information using the register unit 23 in accordance with command information input by the operator from the operation input unit 2l via the bus 22 based on the program stored in the program memory 20. .

In this embodiment, the CPU 17 receives the synchronization signal separated from the video signal VD by the synchronization separation circuit section 25,
A video data input circuit 18 having an analog/digital conversion circuit configuration samples the video signal VD, converts it into video data DATA, and loads the video data into a corresponding memory area of the recognition information memory 19.
The CPU 17 transmits a control signal CONT to the injection machine main body 15 via the control signal input/output circuit section 27,
In this way, as the injection molding machine main body l5 executes each step of the injection molding cycle, the corresponding data processing can be executed. The video signal VD obtained from the television camera l6 is supplied to the monitor 32 via the combining circuit section 31, and as a result, the visual field FI is displayed on the display screen PIC, as shown in FIG.
Displays the video corresponding to the LD. Along with this, the CPU 17 monitors the monitoring area Kl%K2...
By supplying the monitoring film display data KDIS representing . The monitoring nod areas Kl, K2, etc. are superimposed on the display. In this embodiment, the CPU 17 controls the register section 23 (second
The mark shapes to be used in the monitoring areas Kl, K2, etc. and the monitoring are determined by the mark position designation data D2 of the mark position designation data register REG2 and the mark shape designation data D1 of the mark shape designation data register REGI in the figure). Position WPI (X +, V +), P2 (X!
．． 7)... is given to the graphic display control unit 33 as monitoring area display data KDIS,
The graphic display control unit 33 generates a monitoring area display signal VD, which is a video signal that increases the brightness of the display screen PIC at a timing corresponding to the monitoring area display data KDIS.
Generates K. In addition, in this embodiment, the monitoring areas K1, K2...
A plurality of mark data having different external shapes and sizes are prepared as mark shapes, and when an operator designates one of them through the operation input section 21, the designated data can be stored as mark shape designation data D1. ing. The CPU 17 applies brightness information in the monitoring areas K1, K2, . The recognition information memory 19 stores video data D obtained at the timing when the injection molding machine main body 15 is in the state after mold opening and before ejection.
A first frame memory 19A that captures ATA, and then video data DATA obtained by tying after projection.
a second frame memory 19B that takes in the CP
U17 is the first and second frame memory 19A and 19
Among the video data DATA stored in B, the mark position data register REG of the register section 23 (FIG. 2)
The shifter 35A of the monitoring data type control unit 35 and Integrator 3
5B, and outputs the result as monitoring data DWCH to the post-mold opening and pre-ejection monitoring data register REG5 and the post-ejection monitoring data register REG6 of the register section 23 (FIG. 2).
The data are stored as post-mold opening and pre-ejection monitoring data D5 and post-ejection monitoring data D6, respectively. Thus, when the ejection command machine main body l5 makes a normal ejection operation, the value of the monitoring data DWCH at each monitoring point becomes the judgment level consisting of the judgment standard value data D4 stored in the judgment standard value data register REG4 (Fig. 2). On the other hand, if the injection product is not pushed down and remains, the value of the monitoring data DWCH does not change so as to cross the judgment level formed by the judgment reference value data D4. It is possible to judge whether or not the main body 15 of the ejecting product has been properly dropped. The CPU 17 uses the past multiple mold opening, pre-ejection monitoring data D5 and post-ejection monitoring data D as the determination reference value data D4.
6, the intermediate value thereof, ie, 2, is calculated and stored in the judgment criterion small data register REG4 of the register section 23.

(2) Initial setting of the monitoring area The CPU 17 presets the standard product (
In other words, while displaying the image of the master),
The monitoring area Kj (j=1, 2, etc.) is automatically initialized to the optimal position. In other words, the display screen PIG is
As shown in FIG. 4, the area is divided into automatically set sections SECll-SEC34 of 4×3=12 sections in a grid pattern, with 4 sections and 3 sections in the X direction and y direction, respectively. Each automatic setting section SECI 1-SEC34 is formed by arranging 3x3=9 initial setting ranges SET in a grid shape in the X direction and the y direction, and each initial setting range SET is 5 x 5 = 25 in direction and y direction
It consists of an array of sample video data representing the brightness of each pixel. In the monitoring area initial setting mode, the CPU 17 integrates the sample video data included in each initial setting range SET for the nine initial setting ranges SET included in each of the automatic setting sections SECI L SECl2...SEC34. By doing this, the initial setting range brightness information consisting of the integral brightness data of the initial setting range SET is obtained for the timing before and after the ejection of the ejector type aircraft main body l5, and the initial setting range before and after the ejection is determined. The initial setting range SET in which the deviation of the setting range brightness information is maximum (therefore having the brightness information that is easiest to judge) is automatically set as the initial setting range representing each section SECII, SEC12...SEC34, i.e. The representative initial setting range SETpm (shown with diagonal lines in FIG. 4) is determined. In this way, the first stage of brightness information processing is completed, and the CPU
17 is followed by the second step of brightness information processing, which is the representative initial setting range SET? ．． The optimal monitoring position with the brightness information that is easiest to determine is determined from among these.

That is, as shown in FIG. 5, the CPU 17 determines the positions (x=t, y=u), (x=t+-1, y =
u) When the monitoring area Kj is set at (x=t+4, y=u+4), sample video data that falls within the range of the mark shape of the monitoring area Kj (in this example, 3
The monitoring area brightness information is formed by calculating the integrated value of X3 = 9 sample video data), and the deviation of the monitoring area brightness information before and after the projection is maximum at each sample point (therefore, The sample point (having the brightness information that is easiest to judge) is set in the representative initial setting range SE.
The optimal setting position of TPII is determined, and this is stored in the mark position designation data register REG2 of the register section 23 as mark position designation data D2. In this embodiment, the first and second frame memories 19A and 19B of the recognition information memory 19 (FIG. 1) are stored in the pre-projection and post-projection frames whose addresses are the intersections of the sampling points in the X direction and the sampling points in the y direction. Sample video data DATA of each sample point is stored, and the CPU 17 executes the above-mentioned first and second stage brightness information processing using the sample video data. The position of the monitoring area Kj (J=1, 2...) is determined by using the address of the upper left corner point of the mark shape (in the case of Fig. 5, the sample point Qjt+a (x=t, y=u)). The position can be determined by specifying the point, and the value obtained by integrating the video data of the sample point that falls within the range of the mark shape based on the point at the upper left corner is calculated as the monitoring area Kj (j = 1, 2...
・) As monitoring data DWCH consisting of monitoring area brightness information representing the brightness of
Monitoring data after mold opening and before ejection (this is the first
(referred to as second monitoring data) D5 and post-ejection monitoring data (referred to as second monitoring data) D6. In this way, when the operator initializes the monitoring position Qjv- of the monitoring area Kj for the standard product (i.e., master) in advance, the exact same image will be displayed on the video when manufacturing the injection molding product in each injection molding cycle thereafter. When the data is imported as data DATA, the CPU 17 stores the monitoring area brightness information obtained by integrating the monitoring area KJ based on the monitoring position 1iQjtu as appropriate first and second monitoring data D5 and D6 in the register unit 23. can be imported into.

After the object recognition device 14 automatically sets the monitoring areas K1, K2, etc. for the master as described above, the processing procedure shown in FIG. Execute object recognition processing according to the following.

[3] When the object recognition processing CPU 17 enters the object recognition processing operation from step SPI in FIG. It waits for the data D1 to be manually operated, and when the input operation is performed, the input data is taken into the mark shape specification data register REGI. Subsequently, in step SP3, the CPU 17 selects the sample bidet after mold opening and before ejection, which is taken into the first frame memory 19A of the recognition information memory 19. The monitoring data forming unit 35
All automatic configuration sections SECI 1-SEC3 in
4 (Figure 4), find the integral data of the initial setting range SET, and determine the initial setting range SET with the largest deviation of the integral data as the representative initial setting range SET■. 12 thus formed on the display screen PIC (Fig. 4)
One representative initial setting range SETPI is determined for each of the automatic setting sections SEC1 1 to SEC3 4, but the CPU 17 determines each representative initial setting range SET, . Among them, the sample position M Q J Lu where the deviation of the integral data before and after ejection is the largest is determined as the optimum mark position.

As a result, the sample position W Q J t u that could be determined as the optimal mark position is
When the monitoring area Kj is set at all sample positions included in l~SEC34, the deviation of the integral data before and after ejection is the largest, and thus the monitoring area Kj is set at all sample positions included in SEC34.
(2) This means that the monitoring area Kj can be automatically set at the position where the brightness information makes it easiest to determine the area Kj, and therefore the sensitivity of the monitoring operation is highest. In this state, the CPU 17 moves to step SP5 and executes a process of loading the first monitoring data D5 into the post-mold opening and pre-ejection monitoring data register REG5 (FIG. 2) for the monitoring area Kj set at the optimum mark position. After that, in step SP6, a process of loading the second monitoring data D6 into the post-extraction monitoring data register REG6 is executed. In this way, the CPU 17
Test drive to confirm that it can operate normally to produce a standard injection mold product (i.e. master).
The first and second monitoring data D5 and D6 are taken into the register section 23 as initial data until the start of the subsequent injection cycle.

In this way, the automatic initial setting of the monitoring area Kj before the start of the injection molding cycle is completed, and the CPU 17 moves to step SP7, after which the injection molding machine main body 15- injects the injection molding products one by one in the injection molding cycle. In synchronization with the molding operation, when the injection molding machine main body I5 sequentially performs the inter-mold movement and the ejecting operation, a recognition process is executed to determine whether or not the injection molding product has been properly pushed down. At that time, when a positional deviation occurs in the image pattern with respect to the monitoring head areas K1, K2, etc. that are automatically set on the display screen PIC, the monitoring area K1 is automatically set on the display screen PIC.
, K2... monitoring positions P1, P2...
Execute processing to correct the

That is, the CPU 17 centers on the monitoring position PJ- of the monitoring nod area Kj (FIG. 7) specified based on the mark position designation data D2 written in the mark position designation data register REG2 by the processing of steps SP3 and SP4, Said watch position W. 5×5−25 monitoring points Pjxy(X
= m2, m − 1 ... m + 2, y = n - 2
, n-1...n+2), the first and second
Brightness data when the monitoring area Kj is moved to each monitoring position Pjxy from the frame memories 19A and I9B (that is, the integral value of sample data included in the monitoring area Kj)
is obtained by integral calculation in the monitoring data type control unit 35,
Among the calculation results, the monitoring position i11Pjxy with the largest deviation before and after protrusion is determined.

As a result, the deviation data at the mark position where the deviation is maximum is automatically set at the monitoring position 1fQJ in steps SP3 and SP4. ．． This means that the injection molding machine main body 1 in the current injection mold cycle
This means that the injection or molding operation in step 5 matches the brightness data of the monitoring position automatically set for the master, and this also means that the CPU 17 imports information data that can obtain a practically correct recognition result. It means that you were able to do something. On the other hand, the monitoring position Pj. where the deviation data is maximum. ，
is a position that does not match the automatically set monitoring position, this means that the field of view FI of the television camera 16
This means that the image in the LD may be shifted from the initial position.

However, even if such positional deviation occurs, the monitoring nod area K is initially set at the monitoring position where the deviation data is maximum.
It can be considered that the monitoring area Kj has moved to follow the movement of the image part of Kj in the actual ejection type cycle, and therefore, it can be considered that the monitoring area Kj has moved to the extent that there is no risk of causing erroneous recognition in practical use. This means that we were able to capture highly accurate identification information.

Further, in step SP7, the CPU 17 calculates determination reference value data D4 using equation (1) based on the pre-ejection monitoring data D5 and the post-ejection monitoring data D6. Thereafter, the CPU 17 sends a control signal to the human output circuit section 27 to indicate that the injection molding machine main body 15 has reached the timing after mold opening and before ejection.
When it is confirmed by the control signal CONT fetched from
After importing the first monitoring data D5 from the above into the post-mold opening and pre-ejection monitoring data register REG5 via the monitoring data format controller 35, the pre-ejection monitoring data is determined to have normal brightness with respect to the determination reference value in step SP9. Determine whether or not you have the level.

If a positive result is obtained in step SP9, CPU1
7 moves to step SPIO at the timing when the injection molding machine main body 15 performs the ejecting operation, and outputs the second monitoring data D6 from the second frame memory 19B via the monitoring data format section 35 to the monitoring data register REG6. After importing, it is determined in step SPII whether or not the second monitoring data D6 is normal using the determination reference value.

If a positive result is obtained in step SPII, this means that the operation of the ejection molding machine main body l5 in the relevant ejection molding cycle was normal, and at this time, the CPU 17 returns to the above-mentioned step SP7 and performs a new ejection molding cycle. The calculation of the mark position and judgment reference value for the injection molding cycle is repeated.

On the other hand, if a negative result is obtained in step SP9 or SPII, this means that an abnormality has occurred in the ejection cycle of the ejection command machine main body 15, and in this case, the CPU 17 determines that the operator It waits for a process to eliminate the cause of the failure to be performed, and when the process is completed, returns to step SP8 to repeat the acquisition of new monitoring data D5 and D6 and the determination process. The reason why the CPU 17 does not return from step SP12 to step SP7 is that correct monitoring data D5 and D6 cannot be obtained when a defect occurs, so the CPU 17 does not return to step SP7 from step SP12 without changing the monitoring position of the monitoring area Kj and the judgment reference value data D4. This means that it is used as is in the next determination process.

When executing the calculation in step SP7 during the repetitive processing, the CPU 17 uses the monitoring data D5 and D6 newly taken in in steps SP8 and SP9, and thus each time the ejection command machine main body 15 executes the ejection command cycle once. The monitoring position and judgment reference value data D of the monitoring area Kj are
I will update 4.

? According to the above configuration, when automatically initializing the monitoring area Kj for the video on the display screen PIC, the display screen PIC is divided in advance into a plurality of automatic setting sections SEC11 to SEC34, and each automatic setting section SEC l l-SEC
Among the initial setting ranges SET included in SET, SET, . , and the representative initial setting range SET, . By setting the monitoring area Kj to , it is possible to automatically set the monitoring area to a video portion of the monitoring target that has the greatest reaction to an abnormal operation. Therefore, it is possible to realize an object recognition apparatus that can perform object recognition processing with a determination accuracy that is high enough for practical use, and that can further reduce the burden on the operator since no setting operations are required by the operator.

In addition, according to the above-described embodiment, the monitoring area Kj is set at the monitoring position that has the greatest reaction to an abnormal operation among the sample information included in the representative initial setting range SET■. The monitoring area Kj can be set at a monitoring position where the reaction to an abnormal operation is even greater, and an object recognition device with even higher monitoring accuracy can be realized.

[4] Second Embodiment FIG. 8 shows a second embodiment, in which automatic setting sections SEC41 and S are set in advance on the display screen PIC.
The EC42 is set to have an arbitrary size and shape at an arbitrary position on the display screen PIC.

In other words, the first automatic setting section SEC41 has 12 initial settings having a size that includes 5x5=25 sample video data, as in the case of FIG. 4, at approximately the center of the display screen PIC. It has a size that includes the range SET, and the initial setting range SET is in the x direction and y direction.
They are arranged in a 4x3 shape.

Further, the automatic setting section SEC42 is the first automatic setting section S.
It is set at the lower right position of the display screen PIC, away from EC4 1 to the lower right, and has a size of 8 initial setting ranges SET including 5 x 5 = 25 sample video data, and the initial setting. The range SET is in the χ direction and y
They are arranged in a 2x4-like shape. According to the configuration shown in FIG. 8, when there is a concentrated abnormal reaction area in a part of the display screen PIC as the image to be recognized, the automatic setting section is set only for the image area where the reaction is large. By being able to set SEC41 and SEC42, the amount of data to be processed can be significantly reduced compared to the case where automatic setting sections are set over the entire display screen PIC. The overall size can be made smaller. ? 5] Other embodiments (1) In the case of the embodiment of FIG. 4 or FIG. 8, each automatic setting section SECII to SEC34, or SEC41 and SE
C4 2 is the representative initial setting range SET, . When selecting the representative initial setting range SET■? However, as shown in FIG. 9 or FIG.
1 when the deviation of brightness information exceeds a predetermined judgment level in C34 or SEC41 and SEC42.
A plurality of representative initial setting ranges SETPR, not limited to SETPR, are selected, and the plurality of representative initial setting ranges SET, . The same effect as the above case can be obtained even if the optimum position at which the monitoring area Kj should be set is determined for all the sample data included in the sample data.

(2) Object t in Figure 6! In the embodiment of the identification processing procedure, after the representative initial setting range SET■ is determined in step SP3, the optimum mark position is found from among the sample data included in the representative initial setting range SET■ in step SP4. Instead, when the representative initial setting range SETP,I is determined in step SP3, a predetermined specific position (for example, the center position) may be set as the optimum mark position.

In this way, when the representative initial setting range SETPI is determined, the optimal mark position can be determined in conjunction with the determination of the representative initial setting range SETPI, and the separate object recognition processing procedure can be simplified.

Incidentally, if this is done, in the processing procedure of the subsequent injection molding cycle, when executing the process in step SP7, the monitoring area is determined for the video of the monitored object based on the optimal mark position determined in step SP3. In this case, the monitoring area can be moved to follow the part of the video that actually has the greatest reaction, thereby achieving the same practical effect as the above case. be able to. (3) In the case of the embodiments shown in FIGS. 4 and 8, 9 and 10, each automatic setting section SECII to SEC34, S
Individual monitoring area K for EC4 1 and SEC4 2
We have described the case where Kj is moved, but when it becomes necessary to move one of the monitoring areas Kj, the monitoring area of the entire display screen PIC is moved simultaneously using the method described below. You can do it like this. The first method is to perform calculations to find the optimal monitoring position for one monitoring area selected in advance among the plurality of monitoring areas Kj, and as a result, distances ΔX and Δy in the X and y directions are calculated.
If the calculation result is such that the monitoring position is moved by
Make sure to move only the

In this way, when changing the setting positions of a plurality of monitoring areas arranged on the display screen PIC, the change calculation can be further simplified, and in particular, as in the case of FIGS. 8 and 10, Automatically set section S as part of display screen PIC
It is suitable for application in cases where EC41 and SEC42 are set.

In addition, as a second method, the movement amounts ΔX and Δy are calculated by a function whose variable is the brightness information obtained from the monitoring area Kj set on the display screen PIC, and all the The monitoring area Kj is moved. That is, the monitoring area Kj (j=1, 2...
) deviation Bl(...) of the brightness information before and after the projection
When detecting ΔX, Δy), B2 (ΔX, Δy)..., the following formula B=lB+(ΔX, Δy) +l Bt (ΔX, Δy) +... ... (2) Calculate the brightness difference function B, find the moving amounts ΔX and Δy that maximize the brightness difference function B, and move all monitoring areas Kj by the moving amounts ΔX and Δy (j= 1, 2
...) to move the set positions all at once.

In this way, the moving distance ΔX can be adjusted while taking care not to lose the brightness difference information of the monitoring area Kj on the display screen PIC.
By being able to determine and Δy, the display screen P
The monitoring area Kj can be set with sufficient accuracy for practical use as a whole of rC. (4) In the above embodiment, when integrating the sample video data included in the monitoring area Kj, all the sample video data were integrated under a common weighting coefficient. As shown, the weighting coefficient may be changed as necessary depending on the position of the sample video data within the monitoring area Kj of the sample video data according to the characteristics of the recognition target.

Incidentally, in the case of FIG. 11, when trying to detect a video part whose center point has particularly high brightness on the display screen using the monitoring area Kj, the weighting coefficient for the surrounding sample data is set to r1, and A weighting coefficient of "8" is set for the sample data at the center point. In this way, the monitoring area KJ can be set with the position where the brightness at the center position is high as the optimum position for the monitoring area Kj.

(5) In the above-described embodiment, when sample video data is taken in based on the video signal VD of the television camera l6, the data stored and held in the recognition information memory 19 is transferred to the monitoring data format formed by the shifter 35A and the integrator 35B. Although the case has been described in which the deviation data is expressed in the section 35, instead of this, as shown in FIG. After that, the integrated data is supplied to an integrating circuit 44 consisting of an adder circuit 42 and a memory 43 and accumulated in the memory 43, and is then fetched into the CPU 7 via the bus 22. By doing so, the sample data of the monitoring area is processed in real time. It may be possible to obtain.

In the case of the embodiment shown in FIG. 12, the synchronization signal SYNC included in the video signal VD is separated in the synchronization separation circuit 45 and supplied to the timing control circuit 46, and at the same time
Sample data of initial setting range SET (Fig. 4, Fig. 8, Fig. 9, Fig. 10) or monitoring area Kj from PU17
(A timing signal TM for capturing the sample data of FIGS. 5 and 7 is supplied to the timing control circuit 46 via the bus 22.

When the timing control circuit 46 receives the timing signal TM, the timing control circuit 46 performs an analog/
Digital conversion circuit 4l, addition circuit 42 and memory 43
By supplying a drive signal to the memory 43 of the integrating circuit 44, integral data regarding the initial setting range SET and the monitoring area Kj are accumulated. In the structure shown in FIG. 12, by shifting the timing of the conversion start operation in the analog/digital conversion circuit 4l by a number of clocks, the initial setting range SET or the monitoring area Kj can be shifted in the left and right direction. By shifting the timing ξ by H, the initial setting range SET and the monitoring area Kj can be shifted in the vertical direction. According to the structure shown in FIG. 12, when brightness information is taken in from the video signal VD of the television camera 16 as necessary, the brightness information can be taken in real time, so that changes in the object to be recognized can be handled. It is possible to perform recognition processing with good responsiveness. ? Instead of the configuration shown in FIG. 12, as shown in FIG. timing signal T
Even if the video data corresponding to the initial setting range SET and the monitoring area Kj is taken into the integrating circuit 44 by controlling the opening and closing by a gate signal generated in the timing control circuit 46 based on M, the result shown in FIG. You can get the same effect as in the case. In addition, in the embodiments shown in FIGS. 12 and 13, when automatic setting sections SEC4 1 and SEC4 2 of arbitrary size and shape are set at arbitrary positions as described above with reference to FIG. is automatically set for each successive frame @SEC
41, SEC4 2 data may be imported one section at a time. (6) In the case of the embodiment shown in FIG. 5, the representative initial setting range S
When setting the monitoring area Kj at the optimal monitoring position in ET■, the representative initial setting range SET■? Although integral value data was obtained from all 5 x 5 =25 sample video data in the X direction and the y direction, instead of this, for example, as shown in FIGS. 14, 15, and 16, It may be reduced if necessary, and in this way, the same effect as in the above case can be obtained, and the calculation time can be further shortened.

In the case of Fig. 14, the representative initial setting range SETP* is set as a range of 5 x 5 in the X and Y directions with the monitoring position W Q J tu as the center. Monitoring point Q as data to be imported for judgment
Obtain monitoring data from 8 monitoring points arranged in the X and y directions passing through the monitoring point Qjcu, and 8 monitoring points arranged diagonally through the monitoring point Qjcu. Do it like this. In this way, the representative initial setting range SE can be used as the range of monitoring data to be imported.
While ensuring the size of T■ to be 5 x 5, the number of data to be imported is changed from 5 x 5 = 25 to 1 + 8 + 8 =
The number can be reduced to 17, and the calculation time for determining the optimal monitoring position can be reduced accordingly. In addition, in the case of Fig. 15, the representative initial setting range SE T
P* is limited to a 3×3 range centered on the monitoring point Qj1. In this way, compared to the case of Figure 5, where the number of monitoring points to capture data was 5 x 5 = 25, it can be reduced to 3 x 3 = 9, thus making calculations even easier. It can save time. In the case of Fig. 16, compared to the case of Fig. 15, four monitoring points in diagonal directions centering on the monitoring point QjLu are deleted from the monitoring points from which data should be taken. Furthermore, calculation time can be further reduced. (7) In the above-mentioned embodiment, as described above with reference to FIG. 5, the sample point in the upper left corner of the sample points included in the monitoring area Kj is used as the monitoring point of the monitoring area Kj. However, other sample points may be used, and in short, one of the sample points included in the monitoring area Kj may be used.

(8) In the above embodiment, the monitoring position where the deviation of the video information is maximum is determined as the optimal monitoring point, but in reality, the monitoring point is the television camera l6 or the injection molded object to be measured. Since there are variations due to product vibration, positional deviation, etc., calculate the average value of the data of the optimal monitoring point obtained in multiple injection or mold cycles in the past,
The monitoring position represented by the average value may be set as the corrected monitoring position.

Incidentally, when the average value of the optimal monitoring positions is determined, the monitoring position represented by the average value corresponds to the center of the monitoring positions that vary due to measurements.

(Effects of the Invention) As described above, according to the present invention, a representative initial setting range is selected for each automatic setting section including a plurality of initial setting ranges, based on a video signal obtained by imaging a master, By setting the monitoring area at one of the monitoring points included in the representative initial setting range, it is possible to initially set the monitoring area with sufficiently high accuracy for practical use without placing an excessive burden on the operator. .

Therefore, by setting the monitoring area by selecting the monitoring point from which the optimal monitoring data can be obtained from among the monitoring points included in the representative initial setting range, monitoring can be performed with even higher precision in practical use. You can set the area.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the object recognition device according to the present invention, FIG. 2 is a schematic diagram showing the detailed structure of the register section, and FIG. 3 is a diagram showing settings for a video signal display screen. A schematic diagram to explain the monitoring area, Figure 4 is the display page P.
A schematic diagram for explaining the initial setting range SET formed in the IC, the representative initial setting range SET-■, and the automatic setting sections SECII to SEC34, FIG. 5 shows the representative initial setting range SET
■Please explain the method for setting the optimal monitoring point in ■? A schematic diagram, FIG. 6 is a flowchart showing the object recognition processing procedure of the CPU 17 in FIG. 1, and FIG. A schematic diagram for explaining the correction range AR of tPJ ■, FIG. 8 is a schematic diagram showing another way of setting the representative initial setting range, and FIGS. 9 and 10 are similar to those in FIGS. 4 and 8. A schematic diagram showing other embodiments,
FIG. 11 is a schematic diagram for explaining the weighting calculation method during the integral processing of the monitoring nod area Kj, FIGS. 12 and 13 are block diagrams showing other embodiments of FIG. 1, and FIGS. 16th
This figure is a schematic diagram showing another embodiment of FIG. 5 with respect to the representative initial setting range, and FIGS. 17 to 19 are schematic diagrams showing conventional configurations. l4...Object recognition device, 15...Ejection machine body, 16...Television camera, 1
7...CPU, 19...Recognition information memory, 19A, 19B...Frame memory, 23
. . . Register section, 35 . . . Monitoring data forming section.

Claims (2)

[Claims] (1) A monitoring area is set at a predetermined monitoring position on the display screen represented by a video signal obtained by imaging a recognition target with a television camera, and changes in brightness information obtained from the monitoring area are monitored. In the object recognition device that determines whether the recognition target is normal or not by , while the master is displayed on the display screen, obtain brightness information for each of the initial setting ranges included in the automatic setting section, and among the obtained brightness information, brightness information that is easy to determine whether or not it is normal. An object recognition device characterized in that the initial setting range having the above-described initial setting range is determined as a representative initial setting range, and the monitoring area is set in the representative initial setting range. (2) A monitoring area is set at a predetermined monitoring position on the display screen represented by a video signal obtained by capturing an image of the recognition target with a television camera, and changes in brightness information obtained from the monitoring area are monitored. In the object recognition device that determines whether the recognition target is normal or not by , obtain first brightness information for each of the initial setting ranges included in the automatic setting section with the master displayed on the display screen, and determine whether or not the first brightness information is normal. The initial setting range having easy brightness information is determined as the representative initial setting range, and the monitoring areas are sequentially set with the sample points included in the representative initial setting range as the monitoring positions, and the second brightness information is respectively set. Object recognition characterized in that, among each sample point, a sample point having brightness information that makes it easy to determine whether it is normal or not among the second brightness information is determined as the initial setting position of the monitoring area. Device.