JP3222238B2 - Security monitoring device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a security monitoring device using a security camera.

[0002]

2. Description of the Related Art FIG. 1 shows a conventional security monitoring device. The security camera 1 installed toward the security target area supplies image signals obtained by photographing the security target area to the image memory 2 and the reference image memory 3, respectively. The image memory 2 temporarily stores the binary image signal supplied from the security camera 1 in units of one frame, and supplies the stored image signal of one frame to the comparing unit 4.

In the reference image memory 3, image information (a binary image signal for one frame) of a crime prevention target area in a normal state, which is shot by the security camera 1, is stored in advance. The comparing means 4 compares the image signal for one frame supplied from the image memory 2 with the image signal for one frame stored in the reference image memory 3, and the number of mismatches between the two exceeds a predetermined number. In this case, an alarm signal is supplied to the alarm 5.

The alarm device 5 performs, for example, generation of an alarm sound, lighting of an alarm lamp, and the like in accordance with the supplied alarm signal.
As described above, in the conventional security monitoring device, first, the security target area is photographed by the security camera in the normal state, and the image information at this time is stored in the reference image memory in advance. next,
In the surveillance operation, the image information currently being captured by the security camera is compared with the image information stored in the reference image memory. The system was configured to notify the monitoring personnel of the abnormalities that occurred.

Therefore, in the conventional security monitoring apparatus having the above-described configuration, since the reference image information is fixed, the overall characteristics of the image when viewing the security target area change due to changes in sunshine, climate, or the like. In such a case, it is difficult to maintain the monitoring accuracy, which causes an erroneous determination.
In addition, even when there is a small object that should not be regarded as a foreign substance during monitoring, the small substance is regarded as a foreign substance, and there is a problem that a normal monitoring operation cannot be performed.

[0006]

SUMMARY OF THE INVENTION The present invention has been made to solve such a problem, and is intended to be applied to a case where the overall characteristics of an image in a security target area are changed due to a change in sunshine or climate, or It is an object of the present invention to provide a security monitoring device capable of continuing normal monitoring operation even when there is a small object that should not be regarded as a foreign object while viewing the security target area.

[0007]

A security monitoring apparatus according to the present invention comprises a binarizing means for converting an image signal supplied from a security camera into a binary image signal, and accumulating the binary image signal for each frame. Cumulative average calculating means for calculating the cumulative image signal by adding the image signal, calculating the average of the image signal in one frame from the cumulative image signal every predetermined number of times, and outputting the average as a reference image signal; Counting means for comparing the reference image signal to count the number of mismatches with respect to each pixel position in one frame and outputting a count value; a memory for accumulating the count value for each frame; Threshold generation means for generating and outputting an abnormality determination threshold based on the count value accumulated every time a predetermined number of frames are accumulated, and performing abnormality determination based on the count value and the abnormality determination threshold And a cormorant judgment means.

[0008]

According to the present invention, average image information is obtained every predetermined number of times from the cumulative result of the image information captured by the security camera, and is used as reference image information. And the number of mismatches for each pixel position in one frame is counted. At this time, every time the counting result in the image information for a predetermined number of frames is accumulated, a threshold value for abnormality determination is generated based on the accumulated counting result, and the number of mismatches for each pixel position in one frame is calculated. When the threshold value is exceeded, an alarm operation is performed assuming that an abnormality has occurred in the crime prevention target area.

[0009]

Embodiments of the present invention will be described below. FIG. 2 shows the configuration of a security monitoring device according to the present invention. The security camera 1 installed toward the crime prevention target area supplies an image signal obtained by photographing the crime prevention target area to the contour extraction means 6 and the average luminance calculation means 7, respectively. The contour extraction means 6
The outline component of the image is extracted for each frame of the supplied image signal and supplied to the binarization circuit 8. The contour extraction means 6
For example, it is configured by a Sobel operation circuit or the like, and the supplied image signal is differentiated by the Sobel operation circuit to extract an outline component of the image. The average luminance calculation means 7 calculates the luminance distribution of the image for each frame of the supplied image signal,
Further, the average value of the luminance distribution is calculated, and the binarization circuit 8
To supply. The binarizing circuit 8 uses the average value of the luminance distribution calculated by the average luminance calculating means 7 as a threshold value and converts the contour components of the image extracted by the contour extracting means 6 into binary values of logic "1" and "0". It is converted into an image signal and supplied to the image memory 2a. The image memory 2a temporarily stores the image signal supplied from the binarizing circuit 8 in a unit of one frame, and supplies the stored image signal of one frame to the comparing means 4 and the cumulative average operation circuit 10, respectively. The cumulative average operation circuit 10
The image signal of one frame unit sequentially supplied from the image memory 2a is cumulatively added for each pixel position, and the cumulative addition result is binarized using the cumulative number of times as a threshold. The cumulative average operation circuit 10 calculates the average image signal from the cumulative result for each supplied image signal of one frame by the operation described above to generate an average image signal in one frame, and uses this as a reference image memory. 3a. The reference image memory 3 a stores an average image signal for each frame supplied from the cumulative average operation circuit 10, and sends the stored content to the comparing means 4.

The comparing means 4 compares the one-frame image signal supplied from the image memory 2a with the one-frame image signal stored in the reference image memory 3a for each pixel position. Is generated for each pixel position, and is supplied to the data correction means 13 for each pixel position. The data correction unit 13 divides the supplied comparison result signal for each pixel position into predetermined pixel position areas. At this time, if there is a comparison result signal indicating an isolated comparison result in each section, The logic of the comparison result signal is inverted, and the comparison result signal for each pixel position in one frame is corrected and supplied to the counter 14. The counter 14 counts the number of “1” in one frame of the corrected comparison result signal, that is, the number of mismatches in one frame, and supplies the counting result to the signal bus 11 and the mismatch number memory 15, respectively. The mismatch number memory 15 is
The supplied number of inconsistencies is divided into addresses for each frame and stored, and the stored contents are supplied to the threshold value calculation circuit 16. The threshold value calculation circuit 16 calculates an average value and a standard deviation value of the contents stored in the number-of-mismatches memory 15, and further calculates a threshold value A by Expression 1. The threshold value calculation circuit 16 sends the obtained average value, standard deviation value and threshold value A to the signal bus 11.

[0011]

## EQU1 ## Threshold A = mean + standard deviation × K1 (K1 is a coefficient determined empirically.
Note that the cumulative average operation circuit 10 has a finite cumulative operation capacity. For example, when the cumulative number reaches 200, the cumulative addition result is reduced by 1 /.
The second is to prevent data overflow. At this time, the cumulative number of additions is also set to １ ／.

The alarm device 5 generates an alarm sound, turns on an alarm lamp, and the like in accordance with an alarm device operation command supplied from the determination circuit 20. The determination circuit 20 determines the information supplied from the threshold value calculation circuit 16 via the signal bus 11 and supplies a control command to the alarm device 5. Next, the operation of the security monitoring device according to the present invention in the above configuration will be described.

First, the initial operation will be described. An image signal obtained by photographing the security target area with the security camera 1 is sequentially stored in the image memory 2a for each frame via the contour extracting means 6, the average luminance calculating means 7, and the binarizing circuit 8, The stored contents are supplied to the cumulative average operation circuit 10 and the comparison means 4. The cumulative average operation circuit 10 cumulatively adds the image signals sequentially supplied from the image memory 2a for each frame, and uses the average image signal in one frame obtained by averaging the predetermined number of cumulative additions (for example, 20) as a reference. It is supplied to the image memory 3a. The comparing means 4 uses the image signal stored in the reference image memory 3a as reference image information,
One frame of the image signal supplied from the image memory 2a is compared with the reference image information. The counter 14 counts the number of mismatches in one frame from the comparison result, and supplies the counting result to the mismatch number memory 15 and the signal bus 11. The threshold value calculation circuit 16 calculates the average value and the standard deviation value of the number of mismatches in one frame based on the contents stored in the mismatch number memory 15 and sends them to the signal bus 11.

By the above operation, the image signals for 20 frames are supplied from the security camera 1 to the cumulative average operation circuit 10, and at the time when the accumulation operation for 20 frames is completed, the judgment circuit 20 sets the threshold value operation circuit. It is determined whether the average value and the standard deviation value supplied from 16 satisfy Expression 2. At this time, when Expression 2 is satisfied, it is determined that the image signal obtained by the security camera 1 has little disturbance, and the initial operation is ended to shift to the normal monitoring operation.
If not, the number of trials in Equation 2 is increased by 1 and the initial operation as described above is repeatedly executed until Equation 2 is satisfied.

[0015]

## EQU2 ## Standard deviation ≦ (number of trials × K2 + K3) × average value (K2 and K3 are empirically determined coefficients, here, for example, K2 = 0.1 and K3 = 0.5). The operation will be described.

First, the threshold value calculating circuit 16 obtains an average value and a standard deviation value from the contents stored in the number-of-mismatches memory 15 set in the initial operation, and calculates a threshold value A according to Equation 3.

[0017]

Threshold A = Average value + Standard deviation value × 5 Next, an image signal obtained by photographing the crime prevention target area with the security camera 1 is obtained by extracting the contour extraction means 6, the average luminance calculation means 7 and the binary signal The image data is sequentially stored in the image memory 2a frame by frame through the conversion circuit 8, and the stored contents are stored in the cumulative average operation circuit 10a.
And the comparison means 4. Cumulative average operation circuit 10
Accumulatively adds image signals sequentially supplied from the image memory 2a for each frame, and performs a predetermined number of additions (for example, 20
Each time), an average image signal in one frame obtained by averaging with the cumulative number of additions is supplied to the reference image memory 3a. At this time, the contents of the reference image memory 3a are updated by the average image signal supplied from the cumulative average operation circuit 10 at every predetermined number of additions. The comparing means 4 uses the image signal stored in the reference image memory 3a as reference image information,
One frame of the image signal supplied from the image memory 2a is compared with the reference image information. The counter 14 counts the number of mismatches in one frame from the comparison result.

The judgment circuit 20 compares the threshold value A generated as described above with the count value of the counter 14. If the count value of the counter 14 is equal to or less than the threshold value A, it is determined that no foreign matter has entered the crime prevention target area. By making a determination, the normal monitoring operation as described above is continued. At this time, the threshold value calculation circuit 16
The threshold value A is calculated from the content stored in the number-of-mismatches memory 15 as described above, and the value is updated every predetermined number of times (for example, 10 times) of 0.

When the value of the counter 14 exceeds the threshold value A, the judging circuit 20 sets the reference image memory 3a in order to prevent foreign substance information from being mixed into the reference image memory 3a.
In a write-inhibited state and the normal monitoring operation is continued. At this time, if the state in which the value of the counter 14 exceeds the threshold value A continues five times or more (in the number of additions of the cumulative average arithmetic circuit 10), the alarm device 5 is alerted that foreign matter has entered the crime prevention target area. Sends out a device operation command.

In the above embodiment, the image signal is binarized. However, unless the configuration of the comparison means becomes complicated, n
Processing can be performed by converting the value. If the processing speed is allowed to decrease, the reference image is multiplied so that the operation of the cumulative average operation circuit 10 to prevent the overflow is unnecessary. In the contour extraction means 6 of the above embodiment, the Sobel arithmetic circuit is used to pay attention to the amount of change in the component having a high spatial frequency in the image (corresponding to the contour component of the image). A monitoring device that focuses on the amount of change in each frequency component using an operator that extracts the frequency component can also be used.

[0021]

As described above, the security monitoring apparatus according to the present invention compares the image information being captured by the security camera with the reference image information, and when the number of mismatches by this comparison exceeds a predetermined threshold value. In a security monitoring device that performs an alarm operation assuming that an abnormality has occurred in a security target area, the above-described reference image information and the threshold value are automatically updated at predetermined intervals based on image information captured by a security camera. While generating.

Therefore, according to the security monitoring device of the present invention, the reference image information and the threshold value for performing the abnormality determination are generated while being automatically updated based on the image information captured by the security camera. It is possible to monitor only the abnormalities in the security target area itself by following the changes in the image due to the influence of sunlight, climate, etc., and the changes in the image due to the presence of small objects, enabling accurate monitoring operation. Becomes

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a conventional security monitoring device.

FIG. 2 is a configuration diagram of a security monitoring device according to the present invention.

[Description of Signs of Main Parts]

 2a Image memory 3a Reference image memory 4 Comparison means 10 Cumulative average operation circuit 14 Counter 16 Threshold operation circuit 20 Judgment circuit

Claims (1)

(57) [Claims] 1. An image signal supplied from a security camera is
A binarizing means for converting the image signal into a value image signal; calculating a cumulative image signal by cumulatively adding the binary image signal for each frame; A cumulative average calculating means for calculating an average and outputting the average image as a reference image signal; a count for comparing the binary image signal with the reference image signal to count the number of mismatches with respect to each pixel position in one frame and outputting a count value Means, a memory for accumulating the count value for each frame, and a threshold for generating and outputting an abnormality determination threshold based on the count value accumulated each time the memory accumulates the count value for a predetermined number of frames. A crime prevention monitoring device comprising: a generation unit; and a determination unit that performs an abnormality determination based on the count value and the abnormality determination threshold.