JPS63181587A - Invasion monitoring device - Google Patents

Abstract

PURPOSE:To effectively decide an invader by dividing a fine fluctuation such as the fluctuation of a picture, noise into small areas, smoothing them and examining the size of a continued island form fluctuation area even in the case of influences cased by small animals to be removed. CONSTITUTION:A sum and difference signal for every grid obtained by an integrating means 15 (the quantity of fluctuation at every small area) is fed to a binarization means 16 and binarized by a predetermined first threshold t1. Namely, the area having the fluctuation more than the first threshold t1 is defined to be 1, and others to be 0. A fluctuation area arithmetic means 17 obtains the fluctuation area from adjacent grids having the fluctuation of the picture and the number of the grids of the obtained fluctuation area is obtained. In such a way, the fluctuation area consisting of the continuous small areas having large fluctuation can be obtained and this is a candidate of the invaders. A deciding means 18 compares the number of the grids of the inputted fluctuation small area with a second predetermined threshold t2, to decide the presence and the absence of the invader.

Description

DETAILED DESCRIPTION OF THE INVENTION [Objective of the Invention (Field of Industrial Application) The present invention relates to an intrusion monitoring device that monitors illegal intruders using images.

(Prior Art) FIG. 4 shows a conventional intrusion monitoring device that monitors illegal intruders using images. In the figure, 1 is an image input unit such as an ITV camera (industrial television camera), which converts the input image into a corresponding video signal, and an image processing unit 2
send to The image processing section 2 samples the input video signal at appropriate time intervals, performs A/D conversion, and stores it in the image memory 3 as binary image data. Image memory 3
has two images, which are alternately switched and used at each sampling time.

Now, the difference signal calculation section 4 calculates the difference between the image data at two consecutive times in the image memory 3, and sends the image difference signal to the determination section 5. The determination unit 5 calculates the total sum of the input difference signals, and if it is equal to or greater than a predetermined threshold value, determines that there is an intruder, and sends the determination result to the host machine 6. When the host machine 6 receives the intruder presence signal, it performs appropriate processing, such as issuing an alarm.

However, such conventional intrusion monitoring devices may make erroneous judgments due to the influence of noise (for example, trees in the background swaying in the wind) or may erroneously detect small animals. Furthermore, if the threshold value is raised to prevent these phenomena, there is a problem that the real intruder may be overlooked.

(Problems to be Solved by the Invention) As described above, the present invention has the problem of being susceptible to malfunctions due to the influence of disturbances such as noise and small animals (and being unable to reliably identify the real intruder). The purpose of this invention is to provide an intrusion monitoring device that can reliably identify an intruder and is less susceptible to disturbances such as noise and small animals.

[Structure of the Invention] (Means for Solving the Problems) The intrusion monitoring device of the present invention includes an image input means for inputting an image and outputting a corresponding image signal, and an image signal output from the image input means. means for inputting and outputting a difference signal at predetermined time intervals; means for dividing the image into a grid and integrating the difference signal for each grid; means for comparing the integrated difference signal value with a predetermined first threshold value and binarizing it; The presence or absence of an intruder is determined by determining the number of grid points in the variable region obtained by determining the number of grid points in the variable region obtained by the method, and by comparing the number of grid points in the variable region obtained by this means with a second predetermined threshold value. and means for determining.

(Operation) With the above-described configuration, the thickness of the input image is detected, and the presence or absence of an intruder is determined based on the size.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

In FIG. 1, reference numeral 11 denotes an image input means such as an ITV camera (industrial television camera), which converts an input image into a corresponding video signal and sends it to an image processing means 12. The image processing means 12 samples the input video signal at a certain time interval, for example, at an appropriate time interval such as one second, and performs A/D conversion to detect the intruder.
It is stored in the image memory 13 as digitized image data. The image memory 13 has two images, which are alternately switched and used at each sampling time. Now, the difference signal calculation means 14 calculates the difference between the image data at two consecutive times in the image memory 13, and sends the image difference signal to the integration means 15. The integrating means 15 divides the image into grids and integrates the difference signal input for each grid.

That is, first, the image is divided into a grid pattern at predetermined intervals. For example, one area is 10×10 pixels. That is, 32
A grid-like small area of x24 will be created. Next, the sum of the difference signals is calculated for each small region. The sum difference signal for each small area obtained here represents the fluctuation of the image for each small area.

The sum difference signal for each grid (variation amount for each small area) obtained by the integrating means 15 is sent to the binarizing means 16, where it is binarized using a predetermined first threshold t1. In other words, a region with fluctuations greater than or equal to the first threshold t1 is defined as “
The binarized signal binarized for each grid is sent to the fluctuation region calculation means 17.The fluctuation region calculation means 17 receives the input binary value. The variation area is determined from adjacent grids with image variation from the converted signal, and the number of lattices in the obtained variation area is determined.In other words, continuous "1" is calculated for the variation value of each binary area. Find the area of ``.First, search for an appropriate area of ``1,'' and then search for ``1'' among the four small areas that are in contact with each side of that small area. are repeatedly counted one after another in a chain.In this way, it is possible to find a variable region consisting of successive small regions with large fluctuations.This is a candidate for an intruder. The size of the fluctuation region, that is, the number of continuous fluctuation small regions counted above (number of grids)
is sent to the determination means 18. The determining means 18 determines the presence or absence of an intruder by comparing the input grid number of the small variable area with a predetermined second threshold t2.

For example, if the number of grids in the small variable area is equal to or greater than the second threshold t2, it is determined that there is an intruder, and an abnormality signal is sent to one host machine. When the host machine 19 receives the abnormality signal, it performs corresponding processing, such as issuing an alarm.

Further, this will be specifically explained in detail using FIG. 2. The image input from the image input means 11 is A as described above.
/D conversion and then stored in the image memory 13. The image memory 13 is configured, for example, with 320×240 pixels and 8 bits per pixel. This is, for example, a in Figure 2.
, a'. Here, a is an image without an intruder, and a' is an image with an intruder. Next, this 2
The purpose is to find the difference between two images, for example b(1,j) -1a(1,j)-a -(1,j)
It is determined by 11≦1≦3201≦i≦240. This is as shown in Fig. 2b, and the background disappears through this process, making it possible to extract only the intruder. Next, the difference signal image obtained here is divided into a lattice shape as shown in C in FIG. 2, and here, for example, one region of the lattice shape is composed of 10×10 pixels. That is, one image is divided into 32×24 small areas. In addition, in C of FIG. 2, the area is divided into 12×9 small areas for simplicity.

Next, the sum of the difference signal images of each small area is calculated.

That is, in this way, the sum of the difference signal images for each small region is determined, and the result is as shown in d in FIG. Next, for each small region, the sum of the difference signal images is binarized using a first threshold value t1 set in advance. That is, 1≦m≦321≦n≦24 Then, a binarized pattern as shown in e of FIG. 2 is obtained. Note that "0" is not written in e of FIG. 2.

Next, in this binarization pattern, two-dimensionally consecutive “1”
Count the number of patterns. This is because when an image of an intruder is binarized, it is obtained as a two-dimensional continuous island-like pattern as shown in e in Figure 2, so by examining the size of such an island-like pattern, , it is possible to determine whether or not there is an intruder.

Such counting of island-like patterns can be realized, for example, by an algorithm as shown in the flowchart of FIG. The size of the island-like pattern obtained here (i.e., the number of small areas divided into a grid pattern) is compared with a preset second threshold t2, and if it is greater than or equal to the second threshold t2, it is determined that there is an intruder. For example, an abnormality signal is sent to one host machine.

With such a configuration, fine fluctuations such as image fluctuations and noise are smoothed by dividing into small regions, so that the influence can be suppressed to an extent that they are not affected. In addition, the influence of small animals and the like can be sufficiently avoided because the size of a continuous island-like variable region is examined. Therefore, an intruder can be determined reliably, and the system is less susceptible to disturbances such as small animals and noise.

Note that the parameters used in the above embodiments, ie, the size of the small area, the threshold values tl, t2, etc., may be determined in advance through experiments, and may be finely adjusted depending on the installation location.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to provide an intrusion monitoring device that can reliably identify an intruder and is less susceptible to disturbances such as noise and small animals.

[Brief explanation of the drawing]

Figures 1 to 3 are for explaining one embodiment of the present invention. Figure 1 is an overall configuration diagram, Figure 2 is a diagram for explaining the operation, and Figure 3 is an island-like diagram. FIG. 4 is a flowchart showing an algorithm for realizing pattern enumeration, and is a block diagram illustrating a conventional intrusion monitoring device. 11... Image input means, 12... Image processing means, 13... Image memory, 14...
...Differential signal calculation means, 15...Integration means, 1
6... Binarization means, 17... Fluctuation area calculation means, 18... Judgment means, tl...
- First threshold, t2...Second threshold. Applicant's agent Patent attorney Takehiko Suzue Figure 3 Figure 4

Claims (1)

[Claims] Image input means for inputting an image and outputting a corresponding image signal; inputting the image signal output from the image input means and outputting a difference signal at each predetermined time interval. means for dividing the image into grids and integrating the difference signal for each grid; and comparing the difference signal value integrated for each grid by this means with a predetermined first threshold. and means for binarizing the binarized signal for each grid by this means, determining a variation area from adjacent grids with image variation, and calculating the number of grids in the obtained variation area. An intrusion monitoring device characterized by comprising means for determining the presence or absence of an intruder by comparing the number of grids in the variable area determined by this means with a predetermined second threshold.