JP2009064287A - Intruder detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an intruder detector detecting an intruder to a restricted part such as an important facility or a road by using an input image of a monitor camera. <P>SOLUTION: The intruder detector is provided with an imaging means 1 and an image processing means 2. The image processing means 2 is provided with a division change small area setting means 14 for setting a division change small area in a size smaller than a small area obtained by dividing a change small area where a change is detected, an area difference calculating means 15 for differentiating an area between a background image and an image of the corresponding division change small area in an input image and obtaining an area differential value with respect to the division change small area, and a division background candidate small area setting means 16 for judging that the change with the background image cannot be recognized and setting it to be the division change small area when the area differential value is smaller than a threshold which is previously set. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an intruder detection device.

  In order to detect intruders in areas where entry is prohibited by image analysis of surveillance cameras installed at places where access is prohibited such as facilities and roads, an intruder detection device is used by the surveillance camera and image processing unit. The technology to configure is known. In the conventional intruder detection device, the surveillance camera is installed at a position overlooking the prohibited entry area, images the prohibited entry area and its peripheral part, and transmits the image to the image processing unit.

  Then, the image processing unit inputs the image of the surveillance camera, divides the image into appropriate small areas, and corresponds to the background image acquired in advance for each small area and the background image in the input image The correlation calculation with the image of the small area to be detected is performed, the change small area and the background candidate small area in the image are detected based on the temporal change, and the change detection area is determined by combining the adjacent change small areas. The intruder is detected by determining that an intruder exists in the change detection area when the geometric feature amount of the change detection area is within the feature amount range corresponding to the intruder set in advance. The position in the image is specified. An example of such a conventional intruder detection device is disclosed in Patent Document 1 below.

  In addition, in the conventional intruder detection device described above, when an intruder dressed in a small pattern or a monochromatic object or the like invades a portion with a small pattern on the background image, the correspondence between the background image and the input image In order to compensate for the difficulty of detecting changes by correlation calculation of small areas to be detected, changes in the vicinity of the outline of the object are detected in a wide donut shape by using a relatively large size small area, and intruders are detected. The method of doing is also known.

JP 2006-252248 A

  However, in the conventional intruder detection device described in Patent Document 1, when an intruder who is dressed with a small pattern or a monochromatic object or the like invades a portion with a small pattern on the background image, the background image and It may be difficult to detect changes by correlation calculation of corresponding small areas of the input image.

  Further, although the problem of the conventional intruder detection device described in Patent Document 1 has been solved by using a relatively large size small area, a large size object is used for the small area used for change detection. When there are intruders and intruders nearby, as shown in FIG. 8, the change subregions detected by different intruders and intruders are adjacent to each other, and the change subregions adjacent to each other are combined. May be one large small change area (indicated by E in FIG. 8).

  In view of the above, an object of the present invention is to provide an intruder detection device that detects an intruder or the like in an entry-prohibited area such as an important facility or road using an input image of a surveillance camera.

The intruder detection device according to the first invention for solving the above problem is
An imaging means for imaging the forbidden area and the periphery of the forbidden area;
Image processing means for analyzing an image picked up by the image pickup means,
The image processing means includes
An image input means for inputting an image picked up by the image pickup means to the image processing means as an input image;
A small area dividing means for dividing the input image into appropriate small areas;
Correlation calculating means for performing a correlation calculation between a background image acquired in advance and an image of the corresponding small region in the input image for each small region;
Change detection for detecting a change subregion and a background candidate subregion in the input image based on a temporal change between the background image and the background image corresponding to the background image in the input image Means,
A divided change small area setting means for dividing a change small area in which a change is detected and setting a divided change small area having a size smaller than the small area;
A region difference calculation means for performing a region difference between the background image and an image of the corresponding divided change small region in the input image with respect to the divided change small region, and obtaining a region difference value;
When the area difference value is smaller than a preset threshold, it is determined that a change from the background image is not recognized, and a divided background candidate small area setting unit is set as a divided change small area;
Change area combining means for determining a change detection area by combining adjacent change small areas;
Feature amount evaluation means for evaluating whether or not the geometric feature amount of the change detection region is within a feature amount range corresponding to a preset intruder;
When the geometric feature amount of the change detection area is within a preset feature amount range corresponding to the intruder, it is determined that the intruder exists in the change detection area, thereby detecting the intruder and the input. Intruder detection means for specifying a position in the image is provided.

The intruder detection device according to the second invention for solving the above-described problems is
An imaging means for imaging the forbidden area and the periphery of the forbidden area;
Image processing means for analyzing an image picked up by the image pickup means,
The image processing means includes
An image input means for inputting an image picked up by the image pickup means to the image processing means as an input image;
As the first processing means,
A small area dividing means for dividing the input image into appropriate small areas;
Correlation calculating means for performing a correlation calculation between a background image acquired in advance and an image of the corresponding small region in the input image for each small region;
Change detection for detecting a change subregion and a background candidate subregion in the input image based on a temporal change between the background image and the image of the subregion corresponding to the background image in the input image Means,
A divided change small region / divided background candidate small region setting means for setting a divided change small region and a divided background candidate small region obtained by dividing the change small region and the background candidate small region;
As a second processing means,
A small area dividing means for dividing the input image into small areas having the same size as the divided change small area;
For each of the small regions, a region difference calculation means for performing a region difference between the background image and the image of the corresponding small region in the input image to obtain a region difference value;
When the area difference value is smaller than a preset threshold value, a difference background candidate small area is set. When the area difference value is smaller than a preset threshold value, a difference change small area is set as a difference change small area. Area / difference background candidate small area setting means,
Further, the image processing means includes
Find the logical product of the divided change small area and the divided background candidate small area detected by the first processing means and the difference change small area and the difference background candidate small area detected by the second processing means. A small area AND calculation means;
Change point setting means for determining a change point from the result of the logical product;
Change area combining means for determining a change detection area by combining adjacent change points;
Feature amount evaluation means for evaluating whether or not the geometric feature amount of the change detection region is within a feature amount range corresponding to a preset intruder;
When the geometric feature amount of the change detection area is within a preset feature amount range corresponding to the intruder, the intruder is detected by determining that the intruder exists in the change detection area, and the input Intruder detection means for specifying a position in the image is provided.

According to the first invention, even when an intruder who is dressed with a little pattern or a monochromatic object or the like invades a portion with little pattern on the background image, the intruder or the intruder can be detected.
In addition, even when a plurality of intruders and intruders are present in the vicinity, a large change small area is not generated, and therefore it can be detected as separate intruders and intruders.

  According to the second invention, in addition to the effects of the first invention, change detection by correlation calculation and change detection by area difference can be performed in parallel, so that efficient intruder and intruder detection is performed. be able to.

  Various embodiments of the intruder detection device according to the present invention will be described with reference to FIGS. FIG. 1 is a schematic diagram showing a device configuration of an intruder detection device according to the first embodiment of the present invention, and FIG. 2 is a divided background candidate region in the intruder detection device according to the first embodiment of the present invention. FIG. 3 is a schematic diagram showing a device configuration of an intruder detection device according to the second embodiment of the present invention, and FIG. 4 is a division in the intruder detection device according to the second embodiment of the present invention. FIG. 5 is a diagram showing a change small region and a divided background candidate small region, and FIG. 5 is a diagram showing a difference change small region and a difference background candidate small region in the intruder detection device according to the second embodiment of the present invention. 6 is a diagram illustrating an example of a logical product according to the second embodiment of the present invention, and FIG. 7 is a diagram illustrating a change point and a non-change point in the intruder detection device according to the second embodiment of the present invention. It is.

[First Embodiment]
Hereinafter, the intruder detection device according to the first embodiment of the present invention will be described. FIG. 1 is a schematic diagram illustrating a device configuration of an intruder detection device according to the present embodiment. As shown in FIG. 1, the intruder detection device according to the present embodiment includes a monitoring camera 1 (imaging means) installed at a place where entry is prohibited such as a facility or a road, and an input image obtained by imaging with the monitoring camera 1. And an image processing unit 2 that detects an intruder into the restricted entry area.

  The surveillance camera 1 is installed at a position overlooking the restricted entry area, images the prohibited entry area and its peripheral part, and transmits the image to the image processing unit 2 (image processing means). The image processing unit 2 uses the image input unit 10 to input an image captured by the monitoring camera 1 as an input image.

  After the processing in the image input means 10 is finished, the small area dividing means 11 divides the input image into appropriate small areas. After the processing in the small area dividing unit 11 is finished, the correlation calculating unit 12 calculates a background image acquired in advance for each small area and an image of a small area corresponding to the background image in the input image. Perform correlation calculation.

  After the processing in the correlation calculation unit 12 is completed, the change detection unit 13 causes the change subregion in the input image to change based on the temporal change between the background image and the background image in the input image and the corresponding small region image. And a background candidate small region are detected. Here, a small region having a relatively large size is used as a small region for detecting a change in the input image.

  After the process in the change detecting unit 13 is finished, the divided small change region setting unit 14 divides the small change region in which the change is detected and smaller than the small region (indicated by T in FIG. 2) as shown in FIG. A small division change area of size (indicated by t in FIG. 2) is set. After the processing in the division change small region setting unit 14 is completed, the region difference calculation unit 15 performs region difference between the background image and the image of the corresponding division change small region in the input image with respect to the division change small region. Get the difference value.

  After the processing in the region difference calculation unit 15 is finished, when the region difference value is smaller than a preset threshold by the divided background candidate small region setting unit 16 as shown in FIG. It is determined that the divided background candidate small area (indicated by A in FIG. 2) is determined to be not allowed.

  After the process in the divided background candidate small region setting unit 16 is completed, the change region combining unit 17 combines adjacent small change regions to determine a change detection region. After the processing in the change area combining unit 17 is finished, the feature quantity evaluation unit 18 sets the geometric feature quantity (for example, the perimeter or area) of the change detection area within the preset feature quantity range corresponding to the intruder. Evaluate whether there is.

  After the processing in the feature amount evaluation unit 18 is finished, the intruder detection unit 19 enters the change detection region when the geometric feature amount of the change detection region is within the feature amount range corresponding to the intruder set in advance. By determining that there is a person, the intruder is detected and the position in the input image is specified.

  As described above, according to the intruder detection device according to the present embodiment, even when an intruder who is dressed with little pattern or a monochromatic object or the like invades into a portion with little pattern on the background image, Intruders and intruders can be detected. In addition, by providing a divided change small area obtained by dividing the change small area, and verifying again the presence or absence of the change in the background image in the divided change small area, if no change is recognized, the divided background candidate small area, Since the change detection area is determined by the remaining divided change small areas and intruders are detected, even if multiple intruders and intruders exist nearby, they do not become large change small areas, but separate intruders and intruders. Can be detected.

[Second Embodiment]
Hereinafter, an intruder detection device according to a second embodiment of the present invention will be described. FIG. 3 is a schematic diagram illustrating a device configuration of the intruder detection device according to the present embodiment. As shown in FIG. 3, the intruder detection device according to the present embodiment includes a monitoring camera 1 (imaging means) installed at a place where entry is prohibited such as a facility or a road, and an input image obtained by imaging with the monitoring camera 1. And an image processing unit 2 that detects an intruder into the restricted entry area.

  The surveillance camera 1 is installed at a position overlooking the restricted entry area, images the prohibited entry area and its peripheral part, and transmits the image to the image processing unit 2 (image processing means). The image processing unit 2 uses the image input unit 10 to input an image captured by the monitoring camera 1 as an input image. After the processing in the image input unit 10 is completed, the image processing unit 2 performs the first processing unit 2a and the second processing unit 2b.

  In the first processing means 2a, after the processing in the image input means 10 is finished, the small area dividing means 11 divides the input image into appropriate small areas. After the processing in the small area dividing unit 11 is finished, the correlation calculating unit 12 calculates a background image acquired in advance for each small area and an image of a small area corresponding to the background image in the input image. Perform correlation calculation.

  After the processing in the correlation calculation unit 12 is completed, the change detection unit 13 causes the change subregion in the input image to change based on the temporal change between the background image and the background image in the input image and the corresponding small region image. And a background candidate small region are detected.

  After the processing in the change detecting means 13, the divided change small area / division background candidate small area setting means 20 divides the change small area and the background candidate small area as shown in FIG. 4) (indicated by B in FIG. 4) and a divided background candidate small region (indicated by b in FIG. 4). In this way, the image processing means 2 obtains the divided change small area and the divided background candidate small area by the first processing means 2a.

  In the second processing means 2b, after the processing in the image input means 10 is completed, the small area dividing means 11 divides the input image into appropriate small areas. After the processing in the small region dividing unit 11 is completed, the region difference calculating unit 21 divides the input image into small regions having the same size as the divided change small region.

  After the processing in the region difference calculation unit 21 is completed, the difference change detection unit 22 performs region difference between the background image and a small region image corresponding to the background image in the input image for each divided small region. To obtain the area difference value.

  After the processing in the difference change detecting means 22 is completed, the difference change small area / difference background candidate small area setting means 23 determines that the difference is smaller when the area difference value is smaller than a preset threshold value as shown in FIG. A background candidate small region (indicated by “c” in FIG. 5) is used, and when the region difference value is larger than a preset threshold value, it is determined as a difference change small region (indicated by “C” in FIG. 5). In this way, the image processing means 2 obtains the difference change small area and the difference background candidate small area by the second processing means 2b.

  The image processing means 2 includes a division change small area and a divided background candidate small area detected by the small area AND calculation means 24 based on the correlation calculation in the first processing means 2a, and a second processing means 2b. A logical product of the difference change small region detected based on the region difference and the difference background candidate small region is obtained, and the result is projected onto the change detection plane.

  Here, the logical product of the small areas is, for example, when the small area is divided into a divided change small area and a divided background small area, as shown in FIG. 1 ”,“ 0 ”is assigned to the divided background small area, and when the small area is divided into the difference change small area and the divided background small area, as shown in FIG. “1”, “0” is assigned to the difference background candidate small region, and integration is performed in units of regions at the same position to obtain a logical product as shown in FIG.

  After the processing in the small area logical product calculation means 24 is completed, the change point setting means 25 causes the small area at the same position to be a divided change small area and a difference change small area as shown in FIG. c), the area “1”) is a change point (indicated by D in FIG. 7), and the small area at the same position is not the divided change small area and the difference change small area (for example, in FIG. 6C) , “0” area) non-change point (indicated by d in FIG. 7).

  After the processing in the change point setting unit 25 is completed, the change region combining unit 17 combines adjacent small change regions to determine a change detection region. After the processing in the change area combining unit 17 is finished, the feature quantity evaluation unit 18 sets the geometric feature quantity (for example, the perimeter or area) of the change detection area within the preset feature quantity range corresponding to the intruder. Evaluate whether there is.

  After the processing in the feature amount evaluation unit 18 is finished, the intruder detection unit 19 enters the change detection region when the geometric feature amount of the change detection region is within the feature amount range corresponding to the intruder set in advance. By determining that there is a person, the intruder is detected and the position in the input image is specified.

  As described above, according to the intruder detection device according to the present embodiment, even when an intruder who is dressed with little pattern or a monochromatic object or the like invades a portion with a small pattern on the background image, Intruders and intruders can be detected. In addition, by providing a divided change small area obtained by dividing the change small area, and verifying again whether or not there is a change with the background image in the divided change small area, if no change is recognized, as a divided background candidate small area, Since the change detection area is determined by the remaining divided change small areas and intruders are detected, even if multiple intruders and intruders exist nearby, they do not become large change small areas, but separate intruders and intruders. Can be detected.

  The present invention, for example, installs a surveillance camera for monitoring a prohibited entry location, and analyzes an input image obtained by imaging with the surveillance camera, so that a person who enters the prohibited entry location in the input image can be detected. It is possible to apply to the intruder detection apparatus to detect.

The schematic diagram which showed the apparatus structure of the intruder detection apparatus which concerns on the 1st Embodiment of this invention. The figure which showed the division | segmentation background candidate area | region in the intruder detection apparatus which concerns on the 1st Embodiment of this invention. The schematic diagram which showed the apparatus structure of the intruder detection apparatus which concerns on the 2nd Embodiment of this invention. The figure which showed the division | segmentation change small area | region and the division | segmentation background candidate small area | region in the intruder detection apparatus which concerns on the 2nd Embodiment of this invention. The figure which showed the difference change small area | region and the difference background candidate small area | region in the intruder detection apparatus which concerns on the 2nd Embodiment of this invention. The figure which showed the example of the logical product which concerns on the 2nd Embodiment of this invention The figure which showed the change point and non-change point in the intruder detection apparatus which concerns on the 2nd Embodiment of this invention. The figure which showed the change small area in the conventional intruder detection device

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Surveillance camera 2 Image processing part 10 Image input means 11 Small area division means 12 Correlation calculation means 13 Change detection means 14 Division change small area setting means 15 Area difference calculation means 16 Division background candidate small area setting means 17 Change area combination means 18 Feature amount evaluation means 19 Intruder detection means 20 Division change small area / division background candidate small area setting means 21 Area difference calculation means 22 Difference change detection means 23 Difference change small area / difference background candidate small area setting means 24 Small area AND Calculation means 25 Change point setting means

Claims (2)

An imaging means for imaging the forbidden area and the periphery of the forbidden area;
Image processing means for analyzing an image picked up by the image pickup means,
The image processing means includes
An image input means for inputting an image picked up by the image pickup means to the image processing means as an input image;
A small area dividing means for dividing the input image into appropriate small areas;
Correlation calculating means for performing a correlation calculation between a background image acquired in advance and an image of the corresponding small region in the input image for each small region;
Change detection for detecting a change subregion and a background candidate subregion in the input image based on a temporal change between the background image and the background image corresponding to the background image in the input image Means,
A divided change small area setting means for dividing a change small area in which a change is detected and setting a divided change small area having a size smaller than the small area;
A region difference calculation means for performing a region difference between the background image and an image of the corresponding divided change small region in the input image with respect to the divided change small region, and obtaining a region difference value;
When the area difference value is smaller than a preset threshold, it is determined that a change from the background image is not recognized, and a divided background candidate small area setting unit is set as a divided change small area;
Change area combining means for determining a change detection area by combining adjacent change small areas;
Feature amount evaluation means for evaluating whether or not the geometric feature amount of the change detection region is within a feature amount range corresponding to a preset intruder;
When the geometric feature amount of the change detection area is within a preset feature amount range corresponding to the intruder, it is determined that the intruder exists in the change detection area, thereby detecting the intruder and the input. An intruder detection device comprising intruder detection means for specifying a position in an image. An imaging means for imaging the forbidden area and the periphery of the forbidden area;
Image processing means for analyzing an image picked up by the image pickup means,
The image processing means includes
An image input means for inputting an image picked up by the image pickup means to the image processing means as an input image;
As the first processing means,
A small area dividing means for dividing the input image into appropriate small areas;
Correlation calculating means for performing a correlation calculation between a background image acquired in advance and an image of the corresponding small region in the input image for each small region;
Change detection for detecting a change subregion and a background candidate subregion in the input image based on a temporal change between the background image and the background image corresponding to the background image in the input image Means,
A divided change small area / divided background candidate small area setting means for setting a divided change small area and a divided background candidate small area obtained by dividing the change small area and the background candidate small area;
As a second processing means,
A small area dividing means for dividing the input image into small areas having the same size as the divided change small area;
For each of the small regions, a region difference calculation means for performing a region difference between the background image and the image of the corresponding small region in the input image to obtain a region difference value;
When the area difference value is smaller than a preset threshold value, a difference background candidate small area is set. When the area difference value is smaller than a preset threshold value, a difference change small area is set as a difference change small area. Area / difference background candidate small area setting means,
Further, the image processing means includes
Find the logical product of the divided change small area and the divided background candidate small area detected by the first processing means and the difference change small area and the difference background candidate small area detected by the second processing means. A small area AND calculation means;
Change point setting means for determining a change point from the result of the logical product;
Change area combining means for determining a change detection area by combining adjacent change points;
Feature amount evaluation means for evaluating whether or not the geometric feature amount of the change detection region is within a feature amount range corresponding to a preset intruder;
When the geometric feature amount of the change detection area is within a preset feature amount range corresponding to the intruder, it is determined that the intruder exists in the change detection area, thereby detecting the intruder and the input. An intruder detection device comprising intruder detection means for specifying a position in an image.

Images