JPH1063862A - Method for recognizing mobile object and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and device for improving a follow-up ability to a mobile object moving at a high speed by obtaining a video suitable for the detection and recognition of the mobile object for an object to be recognized moving at a high speed. SOLUTION: A camera 108 is set so that the moving passage of an object to be recognized can be photographed. As for synchronous outputs to the camera 108, vertical synchronizing intervals can be arbitrarily set by instructing an output timing controller 106 by a controlling part 103. Also, a video from the camera 108 is stored through an analog/digital converter 110 in a video memory 113, and can be simultaneously stored in a picture memory 120 by setting from the controlling part 103 to a data controller 116. Moreover, a picture can be processed by interrupting the storage in the picture memory 120 while continuing the storage in the video memory 113.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image recognition system and, more particularly, to a method and apparatus suitable for obtaining a video image having an optimum timing in recognizing a high-speed moving object.

[0002]

2. Description of the Related Art When recognizing a high-speed moving object, it is essential to obtain an image suitable for recognizing a target moving object. For that purpose, it is necessary to detect that the moving object has reached a position suitable for recognition. Conventionally, this detection method includes, for example, a method of detecting an intrusion of a moving object by a sensor as described in Japanese Patent Publication No. 62-7588, and taking an image, and a method of detecting whether a target moving object has come by an image processing device. However, here, a description will be given of a case where the image processing apparatus monitors whether a target moving object has arrived. In the conventional method, the timing of monitoring whether or not a target moving object has arrived has been defined by the specifications of a video signal of a camera to be used. For example, if a non-interlaced NTSC camera is used, images are taken at a minimum of about 16.7 ms from the specification of 60 screens per minute, and the image is image-processed. It is possible to monitor whether or not. However, if image processing is to be performed on the entire area that can be captured by the camera, it takes 16.7 ms by itself, and the time for performing the image processing is lost. Therefore, only the first part of the area that can be captured by the camera is used. Captured image processing was performed.

[0003]

In the prior art, for example, when a camera of the NTSC specification is used in a non-interlaced manner, images can be taken only at a minimum interval of about 16.7 ms. Even if the target moving body is moving at a high speed even after capturing the entire area that can be imaged by the camera for recognition after the detection of the object, the vehicle passes through the recognizable area in 16.7 ms. there were.

In this case, even if an attempt is made to use the image used for detection for recognition, the image for detection is a part of the area that can be imaged by the camera. There was a problem that it could not be recognized.

An object of the present invention is to provide an efficient recognition method and apparatus that can follow a moving object that moves at a high speed that cannot be recognized by the conventional method and that is efficient.

[0006]

SUMMARY OF THE INVENTION It is an object of the present invention to output a vertical synchronization interval of a camera at a shorter interval than a video specification of a camera to a camera which allows an external synchronization input, and to set a minimum interval of imaging. By making the camera shorter than the camera's image specifications, the number of times of monitoring per unit time is increased, and the interval from when the target moving object is detected to when the recognition image is picked up is shortened. This is achieved by enabling imaging of a high-speed moving object that passes through a recognizable area at the vertical synchronization interval for imaging. Note that when capturing the recognition video, a synchronous output to the camera is output in accordance with the camera's original video specifications in order to capture the entire area that can be captured by the camera.

SUMMARY OF THE INVENTION It is an object of the present invention to convert a video signal from a camera from analog data to digital data, convert the analog to digital information, store a recognition area in a video memory, and a detection area in an image memory. Saves the first part of the area, performs detection processing of the target moving object in the image memory, and when the target moving object is detected, the information of the recognition area stored in the video memory is read. It is also achieved by transferring the image data to the image memory and performing the recognition process.

[0008]

According to the present invention, it is possible to output an arbitrary vertical synchronization interval of a synchronization signal to be output to a camera capable of externally inputting vertical synchronization defining a starting point of field scanning. Therefore, when capturing only the first part of the entire image-capable area of the camera, more images can be captured per unit time than when using the vertical synchronization interval for capturing the entire image-capable area of the camera.

According to this, for example, the detection area is about 1/3 of the entire area where the camera can capture images, and the vertical synchronization interval of the synchronization signal output to the camera is set to 1/2 of the total area where the camera can capture images. In this case, the detection processing can be performed twice as much as per unit time than when the camera is used at the vertical synchronization interval for imaging all the imageable areas of the camera, and from the imaging of the moving object in the imaging area to the next imaging. The moving distance between them is 1 /
2, which makes it possible to cope with a high-speed moving body twice as fast as the conventional method.

According to the present invention, a video signal from a camera is converted from analog data to digital data, and the information converted from analog to digital can be simultaneously stored in a video memory and an image memory. The video signal from the camera can be continuously converted from analog data to digital data while the image processing is aborted halfway, and the information converted from analog to digital can be stored in the video memory.

According to this, while only the first part of the whole area that can be picked up by the camera for detecting the moving object is stored in the image memory, the image can be picked up by the camera for recognizing the moving object at the same timing. It is possible to save the area in the video memory, and when a moving object is detected, it is possible to transfer the data of all the imageable areas of the camera stored in the video memory to the image memory and perform the recognition process, It is possible to improve the ability to respond to a high-speed moving object as compared with the conventional method.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a schematic block diagram of an image processing apparatus according to the present invention. The image processing device 102 includes a CPU, a memory,
The control unit 103 includes an I / O interface and the like, and the image processing unit 104. Note that the image processing unit 104
A plurality of units may exist under one control unit 103. Next, the installation of the camera 108 will be described with reference to FIG. The camera 108 is installed so that the moving object 201 to be recognized passes through the center or almost the center of the imaging area 203 from above to below the imaging area 203. That is, with respect to the moving object 201 to be recognized, the scanning start point 204 with respect to the imaging area of the camera starts from the direction in which the moving object comes, and the scanning end point 205 with respect to the imaging area of the camera moves in the direction in which the moving object leaves. Install.

Next, the variation of the synchronization interval will be described in detail based on FIG. 4 with reference to FIGS. Here, as an example, a case will be described in which the vertical synchronization interval is set to の of the vertical synchronization interval for imaging the entire imageable area of the camera. First, at 401, the output timing controller sets the vertical synchronization interval to １ ／ of the vertical synchronization interval for imaging the entire imageable area of the camera. The setting is performed by the control unit 1 shown in FIG.
03 is set to the output timing controller 106 through the signal line 105. The output timing controller control unit 301 shown in FIG. 3 of the output timing controller 106 stores the number of horizontal scanning lines corresponding to the 1/2 setting value in the register 308. The calculation of the number of horizontal scanning lines may be performed by the control unit 103 of FIG. 1, the control unit 301 of the output timing controller 106 of FIG. Or any of them. Counter 306
Represents a synchronization signal from the synchronization generation circuit 303 on the signal line 30.
4, the number of horizontal scanning lines from the start of the receiving field is counted, and the count value is transmitted to the control unit 309 through the signal line 305. The control unit 309 includes the counter 30
6 is compared with the value stored in the register 308, and when the counter value from the counter 306 matches the value stored in the register 308, the control line 302
The synchronization generation circuit 303 is notified of the coincidence between the counter value from the counter 306 and the value in the register 308 through. Upon receiving the notification from the output timing controller control unit 301, the synchronization generation circuit 303 immediately stops outputting the synchronization signal of the currently executing field and starts outputting the next field. The synchronization signal output to the camera 108 in this manner is different from the vertical synchronization interval 405 of the synchronization signal 403 of the camera for capturing all the imageable areas of the camera, as indicated by reference numeral 404 in 402 of FIG. The video signal from the camera 108 is as shown at 409 in 407.

Next, the writing to the memory will be described in detail with reference to FIG. The video signal from the camera 108 is input to the input timing controller 1 through a signal line 109.
14 and an analog / digital converter (hereinafter referred to as A / D) 1
10 is input. Input timing controller 114
Then, a synchronization signal is separated from the video signal, a timing signal for converting the video signal from the camera 108 into a pixel is generated, and transmitted to the A / D 110, the video memory, and the data controller through the signal line 112. The A / D 110 converts a video signal from the camera 108 from analog data to digital data in accordance with a timing signal from the input timing controller 114, and transmits it to a video memory 113 and a data controller 116 via a signal line 111. The video memory 113 stores the digital signal from the A / D 110. Also, the data controller 116 follows the setting transmitted from the control unit 103 through the signal line 105, and (1) converts the digital signal from the A / D 110 into the signal line 1
19 and stored in the image memory 120.

(2) The digital signal from the A / D 110 is partially stored in the image memory 120 through the signal line 119.

(3) The contents of the video memory 113 are received through the signal line 115 and stored in the image memory 120 through the signal line 119.

(4) The contents of the video memory 113 are passed through the signal line 115, the contents of the image memory 120 are passed through the signal line 119, or the image operation unit 11
8 is received through a signal line 117 and output to a video output unit 122 through a signal line 121.

(5) The contents of the video memory 113 are received through the signal line 115 or the contents of the image memory 120 are received through the signal line 119, and are output to the image operation section 118 through the signal line 117.

(6) The information from the image operation unit 118 is received through the signal line 117 and stored in the image memory 120 through the signal line 119.

The above operations (1) to (6) are performed, and the operations of (1) and (2) satisfy the function of claim 2.

Next, a description will be given of an embodiment based on vehicle number recognition.
5 to 9 show examples applied to a vehicle number recognition system. FIG. 5 is a schematic configuration diagram of the vehicle number recognition system. Camera 1
08 is stored in the camera unit 506,
Is connected to the image processing device 508 via a signal line in the inside. The camera 108 is installed so that the recognition area 505 and the detection area 504 are as shown in FIG. 5 with respect to the moving direction 503 of the recognition target vehicle 501. The vehicle number recognition algorithm will be described with reference to FIG. When the algorithm START 601 is performed, a video capture 602 of the detection area 504 is performed, a vehicle detection process is performed on the captured image, and a vehicle detection determination 603 is performed. When the result of the vehicle detection determination 603 is not detected, a series of operations from the video capturing 602 is repeated. On the other hand, when the recognition target vehicle 501 is detected as a result of the vehicle detection determination 603, the video of the recognition area 505 is captured, and the vehicle number recognition process 604 is performed on the captured image. When the vehicle number recognition processing 604 is completed, the processing 605 of the system is performed based on the result, and then the processing stop determination 60
6 is performed, and if the result of the processing stop determination 606 is a stop, EN
D607 is reached, the process ends, and if the result of the process stop determination 606 is continued, a series of operations from the video capturing 602 is repeated.

Next, the operation using the algorithm of FIG. 6 will be described with reference to FIG. On the time axis 708, graduations are provided at every vertical synchronization interval 707 for imaging the entire imageable area of the camera. The time elapses in the direction of the arrow on the time axis 708. First, when the algorithm shown in FIG. 6 starts, the detection area 504 is fetched into the image memory 120 at the time of the detection area imaging time 701 as shown in the image memory input image 709. The detection processing is performed on the image memory input image 709 of the image memory 120 in the detection processing time 702. As a result, the recognition target vehicle 5
01 is not detected, the detection area 50 is detected again at the time of the detection area imaging time 703 which is the next imaging timing.
4 is captured into image memory 120 as shown in image memory input image 710. The detection processing is performed on the image memory input image 710 of the image memory 120 in the detection processing time 704.
As a result, the recognition target vehicle 501 is detected and the recognition area imaging time 70, which is the next imaging timing, is performed.
At the time of 5, the recognition area 505 is fetched into the image memory 120 as shown in the image memory input image 711, and the recognition processing is performed for the recognition processing time 706. At this time, if the recognition target vehicle 501 is a low / medium-speed vehicle, the progress as indicated by 712 is considered. At the first imaging timing 713 for the detection processing, the vehicle 501 is not detected because the recognition target vehicle 501 has not reached the detection area 504. At the next imaging timing 714 for detection processing, the recognition target vehicle 5
Since 01 has reached the detection area 504, the vehicle is detected. At the next imaging timing 715 for recognition processing, since the license plate 502 of the recognition target vehicle 501 exists in the recognition area 505, the vehicle number can be recognized. However, when the recognition target vehicle 501 is a high-speed vehicle, 7
As shown in FIG. 16, even when vehicle detection can be performed, the movement amount of the recognition target vehicle 501 is large at the imaging timing 719 for the recognition processing, and the license plate 502 is in the recognition area 505.
The car number cannot be recognized because it does not completely fit inside.

Next, the case where the technical concept of variable synchronization according to the present invention is applied will be described in detail with reference to FIG. Assumptions: The detection area 504 is 40% of the imageable area.

The detection processing time is less than 10% of the vertical synchronization interval for imaging the entire imageable area of the camera.

When the detection area 504 is imaged, the vertical synchronization interval is set to one of the vertical synchronization intervals for imaging the entire imageable area of the camera.
/ 2.

At the time of imaging of the recognition area 505, the vertical synchronization interval is set to the vertical synchronization interval for imaging the entire image-capable area of the camera.

Is assumed. First, when the algorithm shown in FIG. 6 starts, the detection area 504 is fetched into the image memory at the time of the detection area imaging time 801 as shown in the image memory input image 807. Image memory 12
The detection processing is performed on the input image 807 of the image memory 807 in the detection processing time 802, and as a result, no vehicle is detected, so the detection area 504 is stored in the image memory 120 in the time of the detection area imaging time 803. Captured as shown in memory input image 808. The detection process is performed on the contents 810 of the image memory 120 in the detection processing time 804, and as a result, the vehicle is detected and the recognition area 505 is stored in the image memory 120 in the recognition area imaging time 805 which is the next imaging timing. Image memory input image 80
As shown in FIG.
6 hours. At this time, if the recognition target vehicle 501 is a low / medium-speed vehicle, a course shown by 810 is considered. At the first imaging timing 811 for the detection process, the vehicle 501 is not detected because the recognition target vehicle 501 has not reached the detection area 504. At the next imaging timing 812 for the detection process, the recognition target vehicle 501 moves to the detection area 5
04, the vehicle is detected. At the next imaging timing 813 for recognition processing, since the license plate 502 of the recognition target vehicle 501 exists in the recognition area 505, the vehicle number can be recognized. Even when the recognition target vehicle 501 is a high-speed vehicle, as shown at 814, the imaging interval has been reduced to 、, so that the vehicle movement amount during that period is also one.
/ 2, the license plate 502 of the recognition target vehicle 501 is present in the recognition area 505 during the recognition processing, and the vehicle number of a high-speed vehicle, which was impossible with the conventional method, can be recognized.

Finally, the case where claims 2 and 3 of the present invention are applied will be described in detail with reference to FIG. The premise is as follows: When imaging the detection area, data is stored in both the video memory 113 and the image memory 120.

The recognition area 50 is stored in the video memory 113.
5 is saved.

When the data of the detection area 504 is imaged, the storage of the image data in the image memory 120 is stopped.

At the time of detection processing, the first part of the detection area 504 (the one where the car enters) is converted into the area on the road where the car actually passes, and is performed on an area where an area of about 30 cm is deleted. . This is to prevent a problem that the license plate 502 does not enter the recognition area 505 because the detection area 504 and the recognition area 505 are simultaneously imaged in the method of the present invention. Further, the deletion width varies depending on the system applied.

The following assumption is made. First, when the algorithm shown in FIG. 6 starts, the detection area 504 is fetched into the image memory 120 at the time of the imaging time 901 as shown in the image memory input image 908, and at the same time, the recognition area 5
05 is the video memory input image 9 in the video memory 113.
06. The detection processing is performed on the image memory input image 908 of the image memory 120 for the detection processing time 902, and as a result, the recognition target vehicle 501 is detected.
Is not detected, the detection area 504 is stored in the image memory 12 at the time of the imaging time 903 which is the next imaging timing.
At the same time, the recognition area 505 is captured in the video memory 113 as shown in the video memory input image 907 as shown in FIG.
The detection processing is performed on the image memory input image 909 of the image memory 120 for the detection processing time 904. As a result, when the recognition target vehicle 501 is detected, the video memory 1
13 video memory input images 907 are stored in the image memory 120.
Is transferred as shown at 912, and the recognition process is performed in the time 905. At this time, if the recognition target vehicle 501 is a low- / medium-speed vehicle, a course shown by 911 is considered. At the first detection processing timing 912, the vehicle 501 is not detected because the recognition target vehicle 501 has not reached the area below the deletion line 915 of the detection area 504. At the next timing 913 for detection processing, the recognition target vehicle 5
Since 01 has reached the area below the deletion line 915 of the detection area 504, the recognition target vehicle 501 is detected. At the next timing 914 for the recognition process, the license plate 502 of the recognition target vehicle 501 is displayed in the recognition area 50.
5, the vehicle number can be recognized. Even when the recognition target vehicle 501 is a high-speed vehicle, the license plate 502 of the recognition target vehicle 501 exists in the recognition area 505 as shown at 916. It becomes possible.

[0034]

According to the present invention, the vertical synchronization of the synchronization signal of the camera can be set at an arbitrary interval. For example, if the period of the vertical synchronization is reduced to half or less, the detection of the moving object can be performed. The imaging interval is also reduced to about 1/2, and the moving distance of the moving body on the screen is reduced to 1/2.

Further, from the above-mentioned effects, 2
There is an effect that it is possible to follow a moving body that moves at double speed.

According to the present invention, in recognizing a moving object,
Images for detection processing can be simultaneously stored in image memory, and images for recognition processing can be simultaneously stored in video memory.
This has the effect of eliminating the time lag between the video for detection processing and the video for recognition processing.

Further, from the effect that the time lag is eliminated, there is an effect that it is possible to follow a moving object that moves at a higher speed (about twice in the method of the present embodiment) than the conventional method.

[Brief description of the drawings]

FIG. 1 is an example of a configuration diagram of an image recognition system of the present invention.

FIG. 2 shows an example of camera installation conditions of the image recognition system of the present invention.

FIG. 3 is an example of a configuration diagram of an output timing controller of the image processing apparatus of the present invention.

FIG. 4 is a flowchart illustrating an example of a schematic operation according to claim 1 of the present invention.

FIG. 5 is a configuration diagram of a vehicle number recognition system according to an embodiment of the present invention.

FIG. 6 is a flowchart illustrating a schematic operation of the vehicle number recognition system according to the embodiment;

FIG. 7 is an explanatory diagram illustrating a schematic operation when a conventional system is applied to the vehicle number recognition system of the embodiment.

FIG. 8 is a block diagram of a vehicle number recognition system according to an embodiment of the present invention;
FIG. 4 is an explanatory diagram for explaining a schematic operation in a case where is applied.

FIG. 9 is a view showing a vehicle number recognition system according to a second embodiment of the present invention;
FIG. 4 is an explanatory diagram for explaining a schematic operation in a case where the first to third embodiments are applied.

[Explanation of symbols]

101: external device, 102, 508: image processing device, 1
03: control unit, 104: image processing unit, 105, 107,
109, 111, 112, 115, 117, 119, 1
21, 123, 302, 304, 305 ... signal lines,
106: output timing controller, 108: camera, 110: analog / digital converter, 113: video memory, 114: input timing controller, 11
6 Data controller, 118 Image calculation unit, 120
... image memory, 122 ... video output unit, 200 ... recognition target moving body, 202 ... recognition target moving body moving direction, 203 ... imaging area, 204 ... scanning start point for camera imaging area, 205 ... scanning end point for camera imaging area , 3
01: output timing controller control unit, 303: synchronization generation circuit, 306: counter, 307: crystal oscillator,
308... Register, 408... Video output signal for imaging the entire imageable area of the camera, 501.
... license plate, 503 ... moving direction, 504 ... detection area, 505 ... recognition area, 506 ... camera unit,
507: pillar, 701, 703, 801, 803: detection area imaging time, 702, 704, 802, 804, 90
2,904: detection processing time, 705, 805: recognition area imaging time, 706, 806, 905: recognition processing time,
707: vertical synchronization interval for imaging the entire area where the camera can capture an image, 708: time axis, 709, 710, 711, 80
7, 808, 809, 908, 909, 910: image memory input image, 901, 903: imaging time, 906, 9
07: video memory input image, 915: deletion line.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Manabu Araoka 5-2-1, Omika-cho, Hitachi City, Ibaraki Prefecture Inside the Omika Plant of Hitachi, Ltd. (72) Inventor Shinpei Yamanaka 5-chome, Omika-cho, Hitachi City, Ibaraki Prefecture No. 2 In the Omika Plant of Hitachi, Ltd. (72) Inventor Yoshitaka Nagasawa 5-2-1 Omika-cho, Hitachi City, Ibaraki Prefecture Inside Hitachi Process Computer Engineering Co., Ltd. (72) Inventor Takayuki Tsutsumi Hitachi, Ibaraki Prefecture 5-2-1 Omikacho Hitachi Process Computer Engineering Co., Ltd.

Claims (3)

[Claims] 1. A recognition device for recognizing a moving object, comprising: a camera capable of externally inputting a synchronization signal timing defining a starting point of field scanning at an arbitrary timing; and an output for generating the synchronization signal timing. An image processing apparatus having a timing controller, wherein in detecting the presence or absence of a moving object, the synchronization signal timing interval is set to 1
It is performed at intervals shorter than the field, and in recognition of the moving object, the synchronization signal timing interval is set to be longer than the interval at the time of detecting the presence or absence of the moving object, and the synchronization signal timing interval is made variable. Characteristic recognition device for moving objects. 2. A recognition apparatus for recognizing a moving object, comprising: a video memory for temporarily storing video signal data from a camera; an image memory for performing image processing; An image calculation unit that performs use image processing; and a data controller that controls writing to the video memory and the image memory and reading and writing from the image calculation unit to the image memory. The data of the video signal is written in parallel to the video memory and the image memory, and the writing to the image memory is stopped at an arbitrary timing even during the writing of one field, and the image calculation unit An apparatus for recognizing a moving object, which permits reading and writing to a memory. 3. A method for recognizing a moving object using a camera capable of variably adjusting a synchronization interval for scanning, wherein presence / absence of the moving object is detected based on a first synchronization interval. A method for recognizing a moving object, comprising recognizing the contents of the moving object at a long second synchronization interval.