JPH0765173A - Intruding body recognition method - Google Patents

Abstract

PURPOSE:To shorten intervals of input image acquisition and to improve recognition precision by performing a recognition processing when an entirely whole image of an intruding body is photographed as an input image. CONSTITUTION:A background image and the input image which are photographed by a TV camera, etc., are stored in memories 11 and 12. In a difference processing 13, the absolute value of the density difference between the background image and input image from the memories 11 and 12 is calculated to find a difference image. In a binarization processing 14, the found difference image is binarized to obtain a difference binarized image. The maximum variation quantity of an input image photographed before the current input image is stored in a memory 16 and the difference binarized image of the input image when the maximum variation quantity is obtained is stored in a memory 17. In a maximum variation detection processing 15, input images having maximum variation quantities are found by using the information stored in the memories 16 and 17. In a variation area segments processing 18, the input images are integrated into one variation area and in an intruding body recognition processing 19, the intruding body is recognized.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an intruding object recognition method for recognizing an intruding object in an intruding object monitoring system for detecting an intruding object from an image picked up by an image input device such as a television camera (hereinafter referred to as a TV camera). It is about.

[0002]

2. Description of the Related Art Conventionally, as a technique in such a field,
For example, some documents were described in the following documents. References: 1990 IEICE Spring National Convention Proceedings 6D-436, Yoshiro Sakuma, Kiyoshi Ito, Isao Masuda “Detecting Intruding Objects by Difference Between Frames of Differential Image” 6-438 FIG. 2 is a diagram showing a processing procedure of a conventional intruding object recognition method described in the above document. In this intruding object recognition method, a background image taken by the TV camera when no intruding object exists is stored in the background image memory 1, and an input image taken by the TV camera when an intruding object exists is input image memory. When stored in 2, the background image of the memories 1 and 2 and the input image are sent to the difference processing 3. Difference processing 3
Then, the absolute value of the density difference between the background image and the input image is calculated as the difference image. This difference image is binarized 4
Is binarized by an appropriate threshold value, an area in which the gray value of the pixel changes between the input image and the background image (this area is called a changed area) is extracted, and sent to the changed area cutout processing 5. In the change area cutout processing 5, the change areas cut out by the difference processing 3 are collected, and the collection result is sent to the intruding object recognition processing 6. In the intruding object recognition processing 6,
Based on the collection result of the changing area cutout process 5, the presence / absence of an intruding object and the detection of the intruding object, that is, the recognition of the intruding object are performed from the feature amount of the changing area. As a method of determining the presence or absence of an intruding object from the feature amount of the changing region, for example, the presence or absence of an intruding object is determined using the area of the changing region and the circumscribed rectangle as parameters. In other words, an intruding object is determined if the area of the change region is equal to or larger than a certain value and within the range of the aspect ratio of the circumscribing rectangle (vertical length / horizontal length of the circumscribing rectangle).
The processing result of the intruding object recognition processing 6 is output in the output processing 7. When the recognition process for one input image is completed as described above, the same recognition process as described above is performed for the next input image.

[0003]

However, the conventional method of recognizing an intruding object has the following problems and it is difficult to solve them. 3 and 4 are diagrams showing an example of an input image in which a part of the intruding object is photographed by the TV camera, and FIG. 5 is a diagram showing an example of an input image in which the whole of the intruding object is photographed. In the conventional intruding object recognition method, the intruding object recognition processing 6 is always performed when a change occurs in the input image stored in the input image memory 2. for that reason,
As shown in FIG. 3, the intruding object recognition process 6 is performed even when a part of the object intrudes. Since the intruding object recognition processing 6 generally has a large amount of processing, assuming that an input image cannot be acquired while the intruding object recognition processing 6 is being performed, the intruding object recognition for a part of the intruding object as shown in FIG. While the process 6 is being performed, the intruding object passes by in front of the TV camera as shown in FIG. As a result, as shown in FIG. 5, an image of the entire intruding object cannot be obtained. That is, the intruding object recognition processing 6
If the target is the whole image of an intruding object, the invasion of the object will be overlooked. The present invention has a problem that the intruding object cannot be accurately recognized by performing the intruding object recognizing process regardless of whether or not the input image includes the entire image of the intruding object. An object of the present invention is to provide a method for recognizing an intruding object, which solves the problem and executes the recognition process only when an image obtained by capturing the entire image of the intruding object is obtained.

[0004]

In order to solve the above-mentioned problems, a first aspect of the present invention provides a background image photographed by an image input device when an intruding object does not exist, and a background image when the intruding object exists. After obtaining the difference images with the input images sequentially captured by the image input device, the difference image is binarized to obtain the difference binarized image, and the intruding object is detected from the change area of the difference binarized image. The following means are taken in the method of recognizing an intruding object. That is, in the present invention, the change amount of the change area between the background image and the input image is sequentially calculated to obtain the input image having the maximum change amount, and the input image captured before the current input image is obtained. The maximum change amount of the change area in the image is stored in the maximum change amount memory, and the difference binarized image of the input image when the maximum change amount is obtained is stored in the maximum change image memory. Then, the intruding object is recognized from the change area of the differential binarized image stored in the maximum change image memory. According to a second invention, in the same intruding object recognition method as that of the first invention, the position of the change region in the input image currently captured by the image input device,
Recognition of the intruding object is performed by comparing the positions of the change areas in the input image previously captured by the image input device and sequentially calculating the change amount of the change areas between the background image and the input image. The input image captured in the most suitable state is detected, and the intruding object is recognized. At this time, the maximum change amount of the input image captured before the current input image is stored in the maximum change amount memory,
The difference binarized image of the input image when the maximum change amount is obtained is stored in the maximum change image memory, and whether the change region in the previously captured input image is located at the edge of the image or not. Store the information in the flag memory. Then, by referring to the storage information of the flag memory, the intruding object is recognized from the change area of the differential binarized image stored in the maximum change image memory and not located at the edge of the image.

[0005]

According to the first aspect of the present invention, since the intruding object recognition method is configured as described above, the input image with the maximum change amount is detected using the stored information in the maximum change amount memory and the maximum change image memory. As a result, the recognition process is executed only when an image of the entire image of the intruding object is obtained. Second
According to the invention, the input image is detected from the stored information of the maximum change amount memory, the maximum change image memory, and the flag memory by using the position and the change amount of the intruding object in the image,
The recognition process is performed when the entire image of the intruding object is captured as an input image without depending on the moving direction of the intruding object. Therefore, the above problem can be solved.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First Embodiment FIG. 1 is a diagram showing a processing procedure of an intruding object recognition method showing a first embodiment of the present invention, and this processing procedure is shown in FIGS.
The description will be made with reference to the above. 6 is a diagram showing an example of the background image of FIG. 1, FIG. 7 is a diagram showing an example of the input image of FIG. 1, and FIG. 8 is a differential binarized image obtained by the binarization processing 14 of FIG. FIG. 9 is a diagram showing an example, and FIG. 9 is a diagram showing a change in area of the change region in FIG. In the intruding object recognition method of FIG. 1, an image input device (for example, a TV camera) is used, and an image as shown in FIG. 6 when no intruding object exists is stored in the background image memory 11 as a background image. Then, at the same position where the background image was captured, an image as shown in FIG. 7 is captured using the TV camera, and the input image is stored in the input image memory 12. The background image and the input image stored in the background image memory 11 and the input image memory 12 are sent to the difference processing 13.
In the difference processing 13, in order to detect a change from the background image in the input image, the absolute value of the density difference between the background image obtained from the background image memory 11 and the input image obtained from the input image memory 12 is calculated. The difference image is calculated and sent to the binarization processing 14. In the binarization process 14, the difference image obtained in the difference process 13 is binarized with an appropriate threshold value to obtain a difference binarized image ΔI as shown in FIG. The difference binarized image ΔI indicates an area (change area) in which the grayscale value of the pixel has changed between the input image and the background image. That is, in this difference binarized image ΔI, the pixel value of the area that has changed from the background image in the input image is set to 1, and the pixel value of the area that has not changed is set to 0, and the changed area is extracted. become.
In this case, the changed area merely indicates a portion where the density has changed from the background image, and does not necessarily indicate the entire image of the intruding object to be recognized. Therefore, it is assumed that the input image having the maximum amount of change from the background image shows the entire image of the intruding object, and the maximum change detection process 15 is performed on the input image having the maximum amount of change from the background image, that is, the difference binary value. The converted image.

As the amount of change, various scales such as the shape of the changed area and the aspect ratio can be used. For example, when the area ΔS of the change area is selected, when the object enters from the right end of the screen as shown in FIG. 3 and goes out from the left end on the opposite side as shown in FIG. 4, the change area from the background image in the input image The change of the area S shows the distribution as shown in FIG. Here, in the input image of the point C where the area ΔS of the change region is maximum,
It is assumed that the whole image of the intruding object is shown. If an object penetrates from the edge of the screen, as the object penetrates,
The area ΔS of the change region changes as shown in FIG. That is, when a part of the object starts to enter from the left end of the screen as shown in FIG. 3, a change area occurs as shown by point A in FIG. When the object continues to enter, the area ΔS of the change region increases as shown by point B in FIG. When the entire intruding object enters the screen as shown in FIG. 5, the area S of the change region becomes maximum as shown by point C in FIG. If an object continues to enter, the object exits from the left edge of the screen as shown in FIG. 4, and the area S of the change region decreases as shown by point D in FIG. Therefore, if the intruding object is to be accurately recognized, it is necessary to perform the recognition process when the entire image of the object is displayed, that is, when the area S of the change region is maximum as at point C in FIG. In the maximum change detection processing 15 of FIG. 1, the input image when the maximum area of the change region is reached is detected. That is, as shown in FIG.
As the object invades, the area S of the change region increases from point A to point B. The input image is sequentially captured and further processed, so that the area S of the change region decreases (see FIG. 9).
The point of the maximum area of the area S of the change region (point C of FIG. 9) becomes clear by detecting the point D) of FIG. Therefore, the maximum change area area ΔSmax obtained in the input images so far and the difference binarized image ΔImax of the change area at that time are respectively stored in the maximum change amount memory 16 and the maximum change image memory 17 of FIG. Save it.

In the maximum change detection processing 15, the area ΔS of the change area of the current input image is compared with the maximum area ΔSmax of the change area of the input image stored so far in the maximum change amount memory 16. If the area ΔS of the change area of the current input image is larger, the maximum area Δ that is the maximum change amount.
While updating Smax, the maximum change image memory 17
The difference binarized image ΔI of the current input image obtained in the binarization processing 14 is saved. Then, the input image of the next frame is read. On the other hand, the current change area area ΔS
Is smaller, the image of the maximum area ΔSmax up to now represents the entire image of the intruding object, and the difference binarized image ΔImax stored in the maximum change image memory 17 is sent to the change area cutout processing 18. . In the change area cutout processing 18, the cutout change areas as shown in FIG. 8 of the binary image generated by the difference calculation between the background image and the input image in the difference processing 13 are integrated (collected) as one integrated change area. For example, in the changing region cutting processing 18, the changing regions of the differential binarized image ΔI are expanded several times and then contracted the same number of times in order to combine the changing regions representing one intruding object into one. The expansion is to convert a pixel having a value of 0 into a pixel having a value of 1 when there is a pixel having a value of 1 around the pixel having a value of 0. In other words, it expands around the change area by one pixel. Contraction is the opposite, and reduces the area around the change area by one pixel. By such an operation, the change area that seems to represent one intruding object is extracted as one area. Then, the change area thus obtained is surrounded by a circumscribed rectangle, and the area surrounded by the circumscribed rectangle is cut out from the difference binary image ΔI. The intruding object recognition processing 19 recognizes the intruding object from the feature amount of the cut-out area. As a recognition method, for example, the intruding object is recognized using the area of the cut out region and the circumscribed rectangle as parameters. That is, the area of the clipped area is greater than or equal to a certain value, and the aspect ratio of the circumscribed rectangle (vertical length / horizontal length of the circumscribed rectangle)
Those within a certain range are recognized as intruding objects. This recognition result is output in the output processing 20, and the reading processing for the next input image is performed.

FIG. 10 shows the maximum change detection processing 15 in FIG.
The processing contents of the maximum change detection processing 15 will be described with reference to this figure. In step S1 of FIG. 10, the area ΔS of the change region of a certain input image is calculated from the difference binarized image ΔI indicating the change region obtained in the binarization processing 14 of FIG. 1, and the process proceeds to step S2. In step S2, the area ΔSmax of the maximum change area so far stored in the maximum change amount memory 16 is compared with the area ΔS of the change area of the input image. If the area ΔS is larger than the maximum area ΔSmax, point A → point B in FIG.
As described above, assuming that the current input image is in the process of expanding the change region, in step S3, the values of ΔSmax and ΔImax stored in the maximum change amount memory 16 and the maximum change image memory 17 are respectively changed to ΔSmax. , ΔI, and the input image is acquired again in the next step S4.
On the other hand, in step S2, if the area ΔS is smaller than the maximum area ΔSmax, the maximum change amount memory 1
The differential binarized image ΔImax having the largest change area so far stored in 6 can be regarded as the differential binarized image at the point C in FIG. 9. Therefore, in step S5, the maximum difference binarized image ΔImax stored in the maximum change image memory 17 is substituted into the difference binarized image ΔI, and is sent to the change region cutout process 19 in step S6.

As described above, this embodiment has the following advantages. In this embodiment, a maximum change detection process 15, a maximum change amount memory 16, and a maximum change image memory 17 are provided, and when the entire image of an intruding object is captured as an input image, the change area cutout process 18 is entered. The object recognition processing 19 performs recognition processing. Therefore, it is possible to shorten the interval for acquiring the input image and improve the recognition accuracy. However, in this embodiment, FIGS.
There is a problem as shown in. FIG. 11 is a diagram showing an example of an input image when an intruding object is photographed far from a TV camera,
FIG. 12 is a diagram showing an example of an input image when an intruding object is photographed near a TV camera, and FIG. 13 is a diagram showing an example of an input image when the intruding object is out of the image.
In this embodiment, it is assumed that the intruding object is located at substantially the same distance from the TV camera. Therefore, the intruding object is T
When approaching the V camera, the change area gradually becomes large, and when moving away from the TV camera, the change area becomes smaller. That is, whether or not the entire image of the intruding object is included in the image cannot be determined only by the size of the area of the change region. For example, consider the case where an intruding object approaches a TV camera. When it is far from the TV camera, the area of the change region is small as shown in FIG. As approaching the TV camera, the size of the intruding object increases and the area increases, as shown in FIG. As shown in FIG. 13, even if the intruding object gets too close and sticks out of the image, the area is still increasing. As described above, since the area of the change area monotonically increases, the intruding object recognition method based on the maximum value of the changing area may not be able to detect the input image including the entire image of the intruding object and suitable for the recognition processing. . Further, even when the intruding object moves away from the TV camera, the area of the change region monotonically decreases, and thus the input image suitable for the recognition process may not be detected. Therefore, if such a problem may occur, the intruding object recognition method of the second embodiment described below may be used.

Second Embodiment FIG. 14 is a diagram showing a processing procedure of an intruding object recognizing method showing a second embodiment of the present invention, and FIG. 1 showing the first embodiment.
Elements that are the same as the elements inside have the same reference numerals.
In this intruding object recognition method, the maximum change detection process 1 in FIG.
5, a recognition control process 25 and a flag memory 26 are provided. In the recognition control process 25, based on the difference binarized image ΔI obtained in the binarization process 14,
Continue the recognition processing of the intruding object thereafter by comparing the amount of change (area of the changing region) between the input image obtained previously and the input image currently being processed, and comparing the position of the changing region, or A process of interrupting the process and determining whether to obtain the next input process is performed. The flag memory 26 is a memory for storing the content of an information flag indicating whether or not the change area in the difference binarized image ΔImax of the maximum change area obtained in the binarization process 14 is in contact with the edge of the image, For example, flag = 1 is set if the change area is in contact with the edge of the image, and flag = 0 is set otherwise. FIG. 15 is a flowchart of the recognition control process 25 in FIG.
The processing procedure of the intruding object recognition method shown in FIG. 14 will be described with reference to FIG. In FIG. 14, a TV camera is used to capture a background image when there is no intruding object as shown in FIG. 6, and the image is stored in the background image memory 11. The input image as shown in FIG. 7 is captured by the TV camera at the same position where the background image was captured, and the input image is stored in the input image memory 12. In the difference processing 13, in order to detect a change from the background image in the input image, the absolute value of the density difference between the background image and the input image obtained from the background image memory 11 and the input image memory 12 is calculated. , The difference image is obtained and sent to the binarization processing 14. Binarization processing 14
Then, the difference image from the difference process 13 is binarized with an appropriate threshold value to obtain the difference binarized image ΔI. In the difference binarized image ΔI, as in the first embodiment, the pixel value is set to 1 in the area where the density is changed from the background image in the input image, and the pixel value is set to 0 in the area where the density is not changed. Then, the changed area is extracted. Such a difference binarized image ΔI
Is sent to the recognition control processing 25.

In the recognition control processing 25, subsequent intrusion is performed by comparing the amount of change (area of the changing region) between the input image obtained previously and the input image currently being processed, and comparing the position of the changing region. It is determined whether to continue the object recognition process or to interrupt the recognition process and acquire the next input image. The processing content of this recognition control processing 25 is shown in FIG.
This will be described with reference to the flowchart of FIG. First, Fig. 1
In step S11 of 5, it is determined whether or not there is a change region (that is, a region where the pixel value is 1) in the difference binarized image ΔI obtained in the binarization processing 14 of FIG. If not, it is considered that there is no intruding object, and step S12 is performed.
Proceed to and acquire the next input image. If there is a changed area in step S11, the process proceeds to step S13,
It is determined whether the changed area is in contact with the edge of the image. If it is in contact with the edge of the image, it is considered that the intruding object is protruding from the image as shown in FIG. In this case, since necessary information (overall image of the intruding object) is insufficient, accurate results cannot be expected even if the subsequent recognition processing is performed. Therefore, subsequent recognition processing is not performed on the current input image. However,
As shown in FIGS. 11 to 13, when an intruding object approaches the TV camera, the following exceptional processing is performed. That is, when the intruding object approaches the TV camera, the area ΔS of the change region gradually increases. As will be described later, if the whole image of the intruding object is in the image (the change area is not in contact with the edge of the image), the image in which the intruding object is as large as possible (the change area is in contact with the edge of the image The recognition process is performed on an image that does not exist and has the largest amount of change. When the area ΔS of the change area is used as the change amount, this means that the recognition processing is performed when the area ΔS of the change area is maximum as shown in FIG. Therefore, when an intruding object approaches as shown in FIG. 11 to FIG. 13, the optimum image for the recognition processing as shown in FIG. 12 is the position of the change region in FIG. 13 (whether it is in contact with the edge of the image). Must be remembered until Therefore, the information about the previous input image is stored as follows.

That is, the maximum change amount area 16 of the maximum change area obtained up to the previous input image is stored in the maximum change amount memory 16.
The difference binarized image ΔImax of the maximum change area when the area ΔSmax is obtained is shown in the maximum change image memory 17, and it is indicated whether or not the change area in the difference binarized image ΔImax is in contact with the edge of the image. The content of the information flag is stored in the flag memory 26, respectively. Here, flag = 1 if the change area touches the edge of the image, otherwise
Set flag = 0. In step S13 of FIG. 15, if the current change area of the input image is in contact with the edge of the image, the process proceeds to step S14, and the position of the change area in the input image showing the maximum change amount so far is checked. flag ＝
If 1, that is, if the change region of the input image that has shown the maximum change before is in contact with the edge of the image, the recognition process is not performed and the process proceeds to step S15 to acquire the next input image. If flag = 0 in step S14, step S16
Then, it is considered that the entire image of the intruding object was captured in the previous input image, and the binary difference image ΔImax stored in the maximum change image memory 17 is substituted for ΔI. Further, in the current input image, since the change area is in contact with the edge of the image, flag = 1 is set for the next input image. Then, the process proceeds to step S17, and the difference binarized image ΔI is shown in FIG.
4 is sent to the change area cutout processing 18 and the recognition processing is performed in the intruding object recognition processing 19. Step S13 of FIG.
In, if the current change area touches the edge of the image,
In step S18, the area ΔS of this change region is calculated, and in step S19, the value of flag is evaluated as in step S14. If flag = 1, the change area extends to the previous input image, which means that the current input image has become the entire image of the intruding object. Therefore, in step S20, the area ΔS of the current change area is substituted for ΔSmax and stored in the maximum change amount memory 16. Further, the current difference binarized image ΔI is substituted into the difference binarized image ΔImax and stored in the maximum change image memory 17. Further, flag = 0 is stored in the flag memory 26. After that, the process proceeds to step S17, where the current difference binarized image ΔI is sent to the changing region cutout process 18, and the intruding object recognition process 19 performs the recognition process. If flag = 0 in step S19, it means that the changed region is not located at the edge of the image in the current input image and the previous input image. Therefore, as in the first embodiment, the input image having the larger change amount is sent to the change region cutout process 18. That is, in step S21, the current change area area Δ
S is compared with the maximum change amount area ΔSmax so far. If ΔSmax> ΔS, ΔImax is substituted for ΔI in step S22, the process proceeds to step S17, and the difference binarized image ΔI is subjected to the change region cutting process 18
Send to. If ΔSmax <ΔS in step S21, the process proceeds to step S23, the contents of ΔS and ΔI are substituted into ΔSmax and ΔImax, respectively, and the next input image is read from the input image memory 12 in step S24. In the differential binarized image ΔI sent to the change area cutout processing 18 of FIG. 14 by the recognition control processing 25 as described above, the change areas cut out by the difference processing in the change area cutout processing 18 are grouped into one change area. Collected in the area. Then, the intruding object recognition processing 19 recognizes what the areas grouped by the changed area cutout processing 18 are, the recognition result is output by the output processing 20, and the next input image is read.

As described above, the present embodiment is provided with the recognition control processing 25 and the flag memory 26 shown in FIG. 14, and is optimal for recognizing an intruding object by using the position and the amount of change in the image of the intruding object. Since the recognition process is performed on the input image, when the intruding object approaches or moves away from the TV camera, the image capturing the entire image of the intruding object cannot be detected, so that the recognition process is performed accurately. It is possible to solve the problem of the first embodiment that it becomes difficult. That is, since the recognition processing is performed at the stage where the entire image of the intruding object is captured as the input image without depending on the moving direction of the intruding object, the interval of input image acquisition can be shortened and the recognition accuracy can be improved. Can be improved. In addition,
The present invention is not limited to the above-described embodiment. For example, the maximum change detection process 15 of FIG. 1 is processed by a processing procedure other than the flowchart of FIG. 10, or the recognition control process 25 of FIG.
Can be processed in various ways other than the flowchart of FIG. 15, and various modifications are possible.

[0015]

As described in detail above, according to the first aspect of the invention, since the maximum change amount detection processing is performed using the maximum change amount memory and the maximum change image memory, the entire image of the intruding object can be obtained. The recognition process can be performed when the image is captured as an input image. Therefore, it is possible to shorten the interval for acquiring the input image and improve the recognition accuracy.
According to the second invention, the maximum change amount memory, the maximum change image memory, and the flag memory are used, and the input image is detected by the recognition control process using the position and the change amount of the intruding object in the image. Therefore, when the intruding object approaches or moves away from the image input device, it is possible to solve the problem that it is difficult to perform the recognition process with high accuracy because the image of the entire image of the intruding object cannot be detected.
That is, since the recognition process can be performed at the stage where the entire image of the intruding object is captured as the input image without depending on the moving direction of the intruding object, the interval of input image acquisition can be shortened and the recognition accuracy can be improved. Can be further improved.

[Brief description of drawings]

FIG. 1 is a diagram showing a processing procedure of an intruding object recognition method according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a processing procedure of a conventional intruding object recognition method.

FIG. 3 is a diagram showing an example of an input image in which a part of an intruding object is photographed.

FIG. 4 is a diagram showing an example of an input image in which a part of an intruding object is photographed.

FIG. 5 is a diagram showing an example of an input image in which an entire intruding object is photographed.

FIG. 6 is a diagram showing an example of a background image.

FIG. 7 is a diagram showing an example of an input image.

FIG. 8 is a diagram showing an example of a difference binarized image.

FIG. 9 is a diagram showing a change in area of a change region.

10 is a flowchart of a maximum change detection process in FIG.

FIG. 11 is a diagram showing an example of an input image when an intruding object is photographed far from a TV camera.

FIG. 12 is a diagram showing an example of an input image when an intruding object is photographed near a TV camera.

FIG. 13 is a diagram showing an example of an input image when an intruding object extends beyond the image.

FIG. 14 is a diagram showing a processing procedure of an intruding object recognition method according to the second embodiment of the present invention.

15 is a flowchart of a recognition control process in FIG.

[Description of Reference Signs] 11 background image memory 12 input image memory 13 difference processing 14 binarization processing 15 maximum change detection processing 16 maximum change amount memory 17 maximum change image memory 18 change area cutout processing 19 intruding object recognition processing 20 output processing 25 Recognition control process 26 Flag memory

Claims (2)

[Claims] 1. A difference image between a background image photographed by the image input device when no intruding object exists and an input image sequentially photographed by the image input device when the intruding object exists is obtained. After that, the difference image is binarized to obtain a difference binarized image, and in the intruding object recognition method for recognizing the intruding object from the change region of the difference binarized image, between the background image and the input image. The change amount of the change area is sequentially calculated to obtain the input image with the maximum change amount,
The maximum change amount of the change area in the input image captured before the current input image is stored in the maximum change amount memory, and the difference binary image of the input image when the maximum change amount is obtained is changed to the maximum change amount. An intruding object recognizing method characterized by recognizing the intruding object from a change region of the differential binarized image stored in the image memory and stored in the maximum change image memory. 2. A difference image between a background image photographed by the image input device when there is no intruding object and an input image sequentially photographed by the image input device when there is the intruding object is obtained. Then, the difference image is binarized to obtain a difference binarized image, and in the intruding object recognition method for recognizing the intruding object from the change region of the difference binarized image, the input currently captured by the image input device is used. The position of the change area in the image is compared with the position of the change area in the input image previously captured by the image input device, and the amount of change in the change area between the background image and the input image is sequentially calculated. The calculated input image captured in the most suitable state for recognizing the intruding object is detected, and the maximum change in the input image captured before the current input image is detected. Storing the amount in the maximum change amount memory, storing the difference binarized image of the input image when the maximum change amount is obtained in the maximum change image memory, and changing the change area in the previously captured input image. Information on whether the image is located at the edge of the image is stored in the flag memory, the stored information of the flag memory is referred to, and the difference binarized image stored in the maximum change image memory and not located at the edge of the image A method for recognizing an intruding object, characterized by recognizing the intruding object from a change area of