JPH0261793A - Intruder detecting device - Google Patents

Abstract

PURPOSE:To exactly detect the presence and absence of an invader by setting plural mask areas, to which weighting is executed based on difference between a present picture and a referring picture, and detecting which mask area and how much the difference between the present picture and referring picture is generated in. CONSTITUTION:A/D conversion 2 of a picture signal, which is image-picked up by a camera 1, is stored to an input picture memory 3 in a short period. A referring picture memory 4 stores a picture at a normal time not to have abnormality. The difference between the memories 3 and 4 is detected 6, accumulated 12 and binarized 13. Then, the weighting and data conversion 14 is executed to the output of the binarizing means 13 in each mask area and the output is stored in a mask picture memory 5. For example, weight is reduced in order of a door, a wall and a window. The detected difference 6 is binarized 7, compared with the output of the mask picture memory 5 and selected 8. Then, the area, to which a picture element to have a luminance change belongs, is obtained and the degree of the change is counted 9. When a count value goes over a prescribed value, such a condition is decided as the abnormality and warning 11 is outputted. Thus, malfunction due to the movement of an object in the outside of the window is prevented.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Field of Application) The present invention relates to an intruder detection device that detects an intruder into a house or the like using an imaging device such as a TV camera.

[Conventional technology]

Conventional devices of this kind are widely used to detect the presence or absence of an intruder based on a difference image between a current image obtained by imaging a monitoring area and a reference image that is stored in advance to indicate the normal state of the monitoring area. It is used.

In addition, in order to improve the reliability of detection, mask areas indicating a plurality of specific kl regions are set in the monitoring area, for example,
When guarding against a window from outside, set the window part as a warning level ■ and the area around the window as a warning level ■, as shown in Figure 5, and when the target T moves from the warning level ■ to the warning level ■, the target T It has been proposed to determine that a person is an intruder and output an alarm (Japanese Patent Laid-Open No. 62-0869).
Publication No. 90). Furthermore, in order to reduce false alarms, non-warning masks are set as non-warning areas (hatched areas in the diagram) in areas that do not require warning (for example, walls of houses without openings such as windows). did.

In the above conventional example, as shown in Figure 6, in the case of indoor monitoring, passersby (not intruders), cars, etc. from outside the window enter the monitoring area through the window, so the entire window is blocked. A security mask area M2 is set, and a security mask area M1 is set over the entire indoor area, which is not intended. In other words, if a Kugera L-T is present indoors, an alarm will be immediately issued. Now, assume that the target Tl is a passerby passing outside the window, and the target T2 is an intruder who has appeared indoors. The target TI will be located only within the unguarded mask region M2 and will not be considered an intruder. Since the target T2 is standing with its back to the window, it is detected both within the guard mask region M1 and in the non-guard mask region M2. Therefore, the area of the target area 1-T2 is divided, and it is determined whether or not there is an intruder based on the area occupied by the pixel with the luminance change within the warning mask region Ml.

[Problem to be solved by the invention]

In this way, if the presence or absence of an intruder is determined based on the small divided area, there is a possibility that the alarm will be missed. In other words, if a large non-warning mask area M2 is adjacent to a small area between warning mask areas, there is a high possibility that a target that straddles both of the mask areas will be separated and misreported. There was a problem with that. Furthermore, in setting the mask area, a pointing device such as a light pen is used to set the mask area according to the screen. However, when setting a mask area with a complicated shape, it is difficult to set the mask accurately, and mask setting requires know-how. There is a problem in that if the mask settings are sloppy, the false alarm rate increases.

The present invention has been made in view of the above points, and its purpose is to be able to reliably detect an intruder even when a target exists across multiple mask areas. Another object of the present invention is to provide an intruder detection device that allows for easy and accurate mask setting.

[Means to solve the problem]

The present invention includes an imaging means for imaging a monitoring area, an A/D conversion means for A/D converting an image signal output from the imaging means, and a current image output from the A/D conversion means. an input image memory, a reference image memory for storing a reference image indicating a normal state of the monitoring area, a difference means for obtaining a difference image between the current image and the reference image, and a determination for determining the presence or absence of an intruder from the difference image. and an alarm output means for outputting an alarm based on the output of the determination means, in which a plurality of a mask image memory for storing a mask area; a selector means for selecting which mask area stored in the mask image memory a pixel having a luminance change in the difference image belongs; and a mask area selected by the selector means. The determination means includes a plurality of counters that count the number of pixels whose brightness exceeds a threshold value for each mask area, and the determination means performs arithmetic processing using the counter value that exceeds a predetermined value, and calculates the result of the calculation. The presence or absence of an intruder is determined by the above method, and the setting of the mask area is determined by a cumulative image memory that cumulatively stores difference images output from the difference means, and a threshold value for the cumulative image stored in the cumulative image memory. and data conversion means that assigns a level to the data binarized by the binarization means according to the positional relationship of the mask area, so that the data is stored in the mask image memory. It is characterized by the fact that

(Operation) In the present invention, mask areas corresponding to a plurality of specific areas are stored in advance in the monitoring area, and levels indicating the positional relationships between the specific areas indicated by these mask areas are set, and each Among the counters corresponding to the mask area, among the counters whose counter value indicating the luminance change pixel exceeds a predetermined value, based on the positional relationship, the counter corresponding to the mask area indicating the specific area closest to the imaging means The counter values of other counters that exceed a predetermined value are added to the value, and the presence or absence of an intruder is determined based on the added value.Furthermore, when setting the mask area, it is necessary to place a person in the monitoring area in advance. By moving, etc., the brightness changes are accumulated in the cumulative image memory, and when the movement of one mask area is completed, it is binarized by the binarization means, and a level indicating the positional relationship between the specific areas indicated by the mask area is generated. After setting, the level indicating the mask area and positional relationship is stored in the mask image memory.

〔Example〕

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be explained below based on drawings showing one embodiment of the present invention.

FIG. 1 is a block diagram showing one embodiment of the present invention. 1
is an image capturing means such as a TV camera, which captures an image of the monitoring area and obtains an image signal. Reference numeral 2 denotes an A/D conversion means which A/D converts the image signal into a digital signal.

Reference numeral 3 denotes an input image memory, which stores for a short time the current image input from the two-dimensional imaging means every moment. 4 is a reference image memory in which a background image indicating a normal state of the monitoring area is stored in advance. Reference numeral 5 denotes a mask image memory for storing the positions of mask images indicating a plurality of specific areas within the monitoring area.

Here, the plurality of mask areas are leveled (i.e., the target T
This leveling is used to indicate which mask area should be prioritized when the mask area exists across two mask areas. For example, in the examples of FIGS. 2 and 3, the level of the mask area corresponding to windows W1 and W2 is set to ■, the level of the mask area corresponding to partition P is set to ■, and the level of the mask area corresponding to partition P is set to ■, corresponding to the floor, wall, etc. Let the level of the mask area be ■. Here, it is assumed that ■ has the highest priority (lowest in the order of , ■, ■).

In other words, if the target T straddles the window Wl and the wall K, the target T (intruder) is always present indoors. Therefore, the level (2) of the mask area corresponding to the wall K is set higher than the level (2) of the mask area corresponding to the window W1. In addition, if targeter L-T straddles partition P and wall K, target T (
(intruder) should exist on the front side of partition P. Therefore, the level ■ of the mask area corresponding to the partition P is set to the level ■ of the mask area corresponding to the wall K.
It is higher than that.

Reference numeral 6 denotes a difference means for obtaining a difference image in brightness between the current image stored in the input image memory 3 and the reference image stored in the reference image memory 4. 7 is a binarization means that binarizes the difference image using a predetermined threshold value and extracts pixels with a change in brightness. 8 is a selector means consisting of a plurality of gates, which outputs the output of the binarization means 7 and the mask image memory 5.
From the output of , it is determined to which mask area the pixel with the luminance change belongs to which specific area corresponds.

The number of gates is equal to the number of mask regions.

Counter means 9 has a plurality of counters corresponding one-to-one to each gate of the selector means 8, and counts the number of changed pixels for each mask area. Reference numeral 10 denotes a determining means, in which the counter value of the counter corresponding to the mask area with the highest level among the counters whose counter value exceeds a predetermined value set in advance is determined by other lower level masks. The counter value of the counter corresponding to the area is added, and the presence or absence of an intruder is determined based on the added counter value. Here, only counters that exceed a preset value are subject to calculation, regardless of the presence of target T, and counters in masked areas where slight change in pixels has occurred due to noise etc. are subject to calculation. This is to exclude it from

Reference numeral 11 denotes an alarm output means, which issues an alarm when the determination means 10 determines that there is an intruder.

Reference numeral 12 denotes a cumulative image memory, which sequentially accumulates and stores the difference images output from the difference means 6. 13 is a binarization means that binarizes the cumulative image stored in the cumulative image memory 12 using a threshold value.

Here, the control of input to the binarization means 13 of the cumulative image is as follows:
For example, a switch S may be interposed between the accumulated image memo IJ l 2 and the binarization means 13, and the person who sets the mask area may turn on the switch at an appropriate time while looking at the screen, and the accumulated image may be What is necessary is to input it to the binarization means 13. Another method is to provide a change detection function in the difference means 6, and when the change is no longer detected,
What is necessary is to send an on signal to the switch.

Reference numeral 14 denotes data converting means, which converts the output of the binarizing means 13 into data whose level is set in advance for each mask area based on the positional relationship of the mask areas, and the output of the data converting means 14 is a mask image. Memory is human-powered.

Next, the operation of this embodiment will be explained using an example in which the monitoring area and mask image shown in FIG. 2 are set.

Fig. 2 shows a scene in which there are two windows W1 and W2 on the left wall, and a partition P can be seen on the right front, which are used to guard against indoor corners.

For this scene, as mentioned above, mask areas are set in order from the specific area closest to the TV left camera, and for each level, partition P is ■, floor, ceiling, wall K is ■, window W
1. W2 becomes ■. The boundaries of each mask area are indicated by dashed lines.

First, in the mask area setting mode, the mask area is set in advance. For example, when setting the window W2 to level ■, as shown in FIG. To go. When the person M finishes crossing the window W2, the switch S is turned on and the cumulative image at that time is input to the binarization means 13. Binarization means 1
In step 3, the cumulative image is binarized using a predetermined threshold value, and pixels with changes are determined. Furthermore, the level of the position of the mask area is set to ■ in advance in the data conversion means 14, and
A level ■ is assigned to pixels determined to have changed by the value converting means. Then, the pixels assigned the level ■ are stored in the mask image memory 5.

Similarly, other mask areas are set and stored in the mask image memory 5.

By setting the mask area in this manner, the mask area can be easily set simply by actually moving a person within the monitoring area, and moreover, the mask area can be set accurately.

After setting the mask area as described above, the detection mode is entered.

FIG. 4 shows a case where the target T is standing indoors with his back to the window Wl. The target T is divided into a wall and a window W1, and counters C2 and C3 corresponding to the mask area showing the wall and window W1 in the counter means 9 have a wall and a window W1.
A value corresponding to the area of the target T occupying each mask region is shown.

Here, level ■ of the mask area indicating wall K is higher than window W1.
Since the level of the mask area indicating
0, the counter value of the counter C3 is added to the counter value of the counter C2, and the addition result is used to determine the presence or absence of an intruder based on a predetermined threshold value. If the addition result exceeds the threshold value, it is determined that there is an intruder, and the alarm output means 11 issues an alarm. Therefore, even if the target T is divided into mask areas representing a plurality of specific areas, it can be reliably detected.

Further, as shown in FIG. 4, when the target T is visible outside the window W2, the counter value can only be added to the counter C3, so addition between the counters is not performed. what if,
Even if the counter value of the counter C3 exceeds the threshold value, if the mask areas indicating the windows W1 and W2 are set as non-warning mask areas, an alarm will not be output erroneously.

〔Effect of the invention〕

The present invention includes an imaging means for imaging a monitoring area, an A/D conversion means for A/D converting an image signal output from the imaging means, and a storage for storing a current image output from the A/D conversion means. an input image memory, a reference image memory for storing a reference image indicating a normal state of the monitoring area, a difference means for obtaining a difference image between the current image and the reference image, and a determination for determining the presence or absence of an intruder from the difference image. and an alarm output means for outputting an alarm based on the output of the determination means, in which a plurality of a mask image memory for storing a mask area; a selector means for selecting which mask area stored in the mask image memory a pixel having a luminance change in the difference image belongs; and a mask area selected by the selector means. The determination means includes a plurality of counters that count the number of pixels whose brightness exceeds a threshold value for each mask area, and the determination means performs arithmetic processing using the counter value that exceeds a predetermined value, and calculates the result of the calculation. The presence or absence of an intruder is determined by the above method, and the setting of the mask area is determined by a cumulative image memory that cumulatively stores difference images output from the difference means, and a threshold value for the cumulative image stored in the cumulative image memory. and data conversion means that assigns a level to the data binarized by the binarization means according to the positional relationship of the mask area, so that the data is stored in the mask image memory. As a result, we were able to provide an intruder detection device that can reliably detect an intruder even when a target exists across multiple mask areas, and also allows for easy and accurate mask setting. .

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the zero skin development, FIGS. 2 to 5 are schematic diagrams of screens for explaining the operation related to the above, and FIGS. 6 and 7 are conventional examples. FIG. 2 is a schematic diagram of a screen for explaining the operation related to FIG. 1.-↑layer image means 2-.A/D conversion means 3-
Human image memory 4.--Reference image memory 5--Mask image memory 6--Differencing means 7--Binarization means 8--Selector means 9--Counter means 10--Determination means
11-Alarm output means 12--cumulative image memo IJ
13-binarization means 14-data conversion means s-switch patent applicant Matsushita Electric Works Co., Ltd.

Claims (1)

[Claims] (1) An imaging means for imaging a monitoring area, an A/D conversion means for A/D converting the image signal output from the imaging means, and an input for storing the current image output from the A/D conversion means. an image memory, a reference image memory for storing a reference image indicating a normal state of the monitoring area, a difference means for obtaining a difference image between the current image and the reference image, and a determination means for determining the presence or absence of an intruder from the difference image. and an alarm output means for outputting an alarm based on the output of the determination means, the intruder detection device comprising a plurality of masks for pre-dividing specific areas of a surveillance image and for distinguishing mutual relationships between the specific areas. a mask image memory for storing a region; a selector means for selecting which mask region stored in the mask image memory a pixel having a luminance change in the difference image belongs; and a mask region selected by the selector means. A plurality of counters are provided for counting the number of pixels whose brightness exceeds a threshold value for each mask area, and the determination means performs arithmetic processing using the counter value that exceeds a predetermined value, and based on the result of this calculation, The presence or absence of an intruder is determined, and the setting of the mask area is determined by a cumulative image memory that cumulatively stores the difference images output from the difference means, and a threshold value for the cumulative image stored in the cumulative image memory. The data is stored in the mask image memory by a binarization means for binarizing the data and a data conversion means for assigning a level to the data binarized by the binarization means according to the positional relationship of the mask area. An intruder detection device characterized by: