JP2667973B2 - Mobile 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 moving object monitoring apparatus for extracting only a necessary moving object image from a television camera image, making a judgment, and detecting the appearance of a monitoring object.

[0002]

2. Description of the Related Art In general, in order to save labor for security and guarding by a person, monitoring by a television camera is automated. FIG. 2 shows a configuration example of a conventional mobile monitoring apparatus used for such a purpose.

In FIG. 2, an image signal picked up by a television camera 21 is digitized by an image input device 22 and then input to a host computer (personal computer) 23. In the host computer 23, for the digitized image data, for example, each pixel is input as horizontal 240 × vertical 224 pixel data having 8-bit luminance data, and this is used as a base image.

Next, the latest image is input in the same manner as the base image, and the difference between each pixel corresponding to the latest image and the base image is calculated. An image constituted by the absolute value is a moving image. This moving image shows a moving part of the latest image compared to the base image.

In order to obtain a moving image from the latest image that follows, by subtracting the pixel of the frame immediately before the frame in which the difference first occurs, only the background that does not move is subtracted, and only the moving object in the latest image is subtracted. Image is obtained. This moving image was displayed, or the appearance was detected and an alarm was issued.

[0006]

As described above, the conventional moving object monitoring apparatus simply extracts the moving image from the difference between the frames of the television camera image, detects the moving object, and issues an alarm. There is a drawback in that even a moving object that does not need to be monitored due to shaking of branches, movement of small animals, or the like is detected, and so-called false alarms are large and impractical. An object of the present invention is to solve such a drawback and to provide a practical mobile monitoring apparatus which does not cause a false alarm.

[0007]

SUMMARY OF THE INVENTION In order to achieve the above-mentioned object, a moving object monitoring apparatus according to the present invention, when extracting an image of a moving object from a television camera image, has a tree moving periodically within a limited range. In order to remove moving body images that do not require monitoring, such as the shaking of branches, the latest frames for each part of the image are removed.
From the level of the moving image of the frame, the moving image of a plurality of predetermined frames following the latest frame is displayed .
The maximum value of the part
A moving image extracting means for extracting a moving image composed only of portions , and further extracting the remaining extracted images for each image group.
The size that is considered according to the screen position, and within the circumscribed rectangle
It has a moving body determining means for determining whether or not it is a moving body to be monitored such as a person or a vehicle from the area ratio or the like.

[0008]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention. As shown in FIG. 1, in the embodiment of the present invention, an image captured by a television camera 11 is digitized by an image input device 12, and then input to a host computer 13 (in this embodiment, a personal computer is used). The host computer includes a moving image extracting unit 14 including a program for removing unnecessary moving images, and a movement determining unit 15 that determines a moving object to be monitored based on the extraction result.
Reference numeral 16 is a memory used for accumulating past base images used by the moving image extraction means 14 or for storing reference data or the like for the determination by the moving body determination means 15. Also one
7 is a display device for monitoring image input,
Reference numeral 18 denotes an alarm generating device that generates a sound or a light to draw a person's attention according to a signal output from the host computer 13 based on the determination result of the moving body determining unit 15.

FIG. 3 is a flow chart showing an operation procedure of the host computer 13 which is a feature of the embodiment of the present invention. In FIG. 3, steps 31 to 35 correspond to the moving image extracting unit 1.
Steps 36 to 42 are steps performed by the moving object determination unit 15.

The host computer 13 executes step 31
Is converted into, for example, horizontal 240 × vertical 22 pixels each having 8-bit luminance data.
Data of four pixels is input as a base image.

Next, in step 32, the latest image is input in the same manner as the base image.

Next, in step 33, the latest image and
Calculate the absolute value of the difference between the base images. This result is the same as the moving image obtained by the conventional technique, and the moving image at this stage is hereinafter referred to as a difference image.

Next, in step 34, the difference image is divided, and 4 × 4 pixels are set as one block, and the average value in the block is calculated. Each of the original 4 × 4 pixels is replaced with one pixel for each block, and the luminance data of the newly replaced pixel is an average value (8-bit configuration) in each block. The image formed by the new pixels thus created is hereinafter referred to as a processed image A. By this data processing, the original image composed of 240 horizontal pixels × 224 vertical pixels has 60 horizontal pixels × 56 vertical pixels, which means that the pixel data has been compressed to 1/16. Then, this processed image A
Is stored in the memory 16 and read out and used whenever necessary.

[0014] Next from the latest processing image A ₀ in step 35, the previous pre-count defined, for example, the pixel data obtained as compared with each relative site of 10 times of the processed image A _1, A ₂ to A ₁₀ The maximum value is subtracted, and as a result, the processed image B is obtained assuming that everything below zero is zero. The processed image B is different from the processed image A in that an image of a moving object that periodically moves in a limited range such as a tree branch swaying in the wind is removed.

For example, a processed image A of a tree as shown in FIG.
At a certain point as shown in FIG.
At another point in time, the image is like C, like b. Therefore, these processed images A _1, A _2, A ₃ ...
Is accumulated, for example, ten times, an image having the maximum value (hereinafter, referred to as a maximum value image) is as shown by a solid line in FIG. When the maximum image of the past 10 times is subtracted from the latest processed image A ₀ (if it becomes negative, it is set to zero), as shown in FIG. 6, the image due to the swing of the tree branch is removed.

On the other hand, since the latest processed image A (shown in FIG. 7) does not overlap with the past processed image (shown in FIG. 8) in the motion of the person or the vehicle, the moving image of the person or the vehicle is shown in FIG. As such, it is not affected by the process of subtracting this maximum value.

Next, in step 36, the processed image B is labeled. Here, the labeling process means that when two pixels having the same value in an image are adjacent to each other including the front, rear, left, right, and oblique directions, the pixels are 4 pixels.
It is said that they are connected (in the case of adjacent front, rear, left and right) or 8 connected (in the case of adjoining including front, rear, left and right and diagonally). And FIG.
As shown in FIG. 1, a process of assigning a label having a different name to each connected component is called a labeling process.

In the labeling process in step 36, a labeling threshold is set, and labeling is performed while checking the connection of pixels only when the pixel data value of the processed image B is larger than the set threshold value. Image binarization (blackening without intermediate color) processing and labeling processing are performed simultaneously to reduce the processing time.

FIG. 10 shows an example of a binarized image, and FIG. 11 shows a result of labeling the binarized image of FIG. 10 (hereinafter referred to as a labeled image).

In step 36, a rectangle circumscribing the labeling image (hereinafter referred to as a circumscribed rectangle) is obtained, and the center coordinates and the vertical and horizontal sizes of the circumscribed rectangle are calculated. FIG. 12 shows the relationship between binarized pixels (large black dots in the figure) and the circumscribed rectangle 50. When the size of the circumscribed rectangle exceeds a predetermined set value (for example, a size that occupies one third of the entire screen), the labeling threshold is too low, and each connected component becomes too large. Therefore, the labeling threshold is automatically increased and the labeling process including the binarization process is performed again.

Next, in step 37, the position of the center coordinates of the circumscribed rectangle, the vertical and horizontal dimensions, and the ratio of the area of the circumscribed rectangle to the number of pixels equal to or larger than the labeling threshold (hereinafter referred to as area ratio) are determined.

Next, in steps 38 to 40, it is judged whether or not the mobile object is a monitored object. In step 38, it is determined whether the center coordinates of the circumscribed rectangle are outside the non-determination area. The non-judgment area is an area set on the screen, and if the center coordinates of the circumscribing rectangle are in this area, it is not judged as a monitoring target. This undetermined area does not need to be monitored on the screen,
In addition, it is set in the area where an image causing false alarm is likely to occur.

In step 39, it is determined whether or not the circumscribed rectangle is larger than the rectangle serving as a criterion. The criterion rectangle is a rectangle to be monitored if the circumscribed rectangle of a certain labeling image is larger than its size, and is set in advance. The setting method is, for example, to divide the screen into eight in the vertical direction, and in a normal camera installation situation, the distance from the camera to the moving object is the longest, and therefore the image of the moving object is the smallest at the top. And the distance from the camera to the moving object is the shortest, so the image of the moving object is the largest, and the judgment reference rectangle is set at the bottom, and the judgment reference rectangle in the other parts is the set value of these two. , For example, automatically calculated by proportional distribution. The determination is made based on which of the eight blocks divided in the vertical direction the center coordinate of the circumscribed rectangle is based on the determination reference rectangle in that block. As described above, an appropriate determination reference rectangle is set in accordance with a change in the size of the moving object image due to the distance from the television camera.

In step 40, it is determined whether the area ratio is larger than the threshold value of the area ratio. The area ratio is a ratio of labeled pixels whose pixel data has a value equal to or larger than a threshold value in the circumscribed rectangle, and as shown in FIG. The proportion (area ratio) occupying the circumscribed rectangle 51 is high, and the ratio (area ratio) of the hatched portion of the pixel occupying the circumscribed rectangle 52 is small as shown in FIG. Therefore, by setting a threshold for the area ratio,
It is possible to distinguish an image caused by movement of a person or a vehicle and an image caused by shaking of a tree branch.

When all of the determinations in steps 38 to 40 are negative and the conditions to be monitored are satisfied, an alarm is issued in step 41 and the process returns to step 32. If it is determined that it is not the monitoring target by any of the steps, step 42
In step 43, the latest image is stored as a base image in step 43, and the process returns to step 32. After returning to step 32, steps 32 to 4
1, or the procedure of steps 32-43 is repeated.

It should be noted that the number of pixels is 60 × 56 (3360 pixels) as a result of the processing in step 34, and the number of pixels after digitization of a normal television image is 512 × 512 (262, 144).
Processing is performed by reducing the amount of data to about 1/78 compared with (pixels).

This is because the purpose of the apparatus is to detect a moving object in which pixels are collected as an image to some extent, and it is not necessary to know the fine state of the moving object. In this way, redundant data for the monitoring process is compressed to simplify the subsequent process. By compressing this amount of data, processing by a personal computer becomes possible and the processing speed is further increased. If the details of the image are required, only the circumscribed rectangle portion of the latest image may be cut out from the coordinates of the circumscribed rectangle of the labeling image in the process of step 36.

Unnecessary moving object images due to the swaying of the branches of the tree are removed in the processing of step 35, and are excluded in the judgments of steps 38 to 40 together with the movement of the small animal. Is very small.

[0029]

As described above, the moving object monitoring apparatus of the present invention, when extracting an image of a moving object, uses various parts of a moving image.
For each of the positions, the maximum value of the pixel data of the moving image in the past several times is subtracted, and the period is within a limited range. <br/> An image of a moving object that does not need to be monitored, such as a tree branch swaying. This has the effect of eliminating the effect of. Further, after the above-mentioned processing, the extracted images that remain further are grouped by image group.
Size that is considered according to the screen position and the circumscribed rectangle
Then the area ratio of the inner or the like, human, by a determination of whether the moving object monitored vehicle or the like, only when it is determined that the monitored, since issuing an alarm, small animals and small
There is an effect that the probability of a false report due to a branch or the like can be extremely reduced.

[Brief description of the drawings]

FIG. 1 is a structural diagram showing an embodiment of the present invention.

FIG. 2 is a configuration system diagram showing an example of a conventional movement monitoring device.

FIG. 3 is a flowchart illustrating an example of a procedure of an operation of the host computer of the embodiment in FIG. 1;

FIG. 4

FIG. 6 is an explanatory diagram showing an operation principle regarding swinging of a tree branch in step 35 of FIG. 3;

FIG. 7

FIG. 9 is an explanatory diagram showing an operation principle regarding movement of a human body in step 35 of FIG. 3;

FIG. 10

FIG. 11 is an explanatory diagram showing an operation principle of labeling in step S36 of FIG. 3;

FIG. 12 is an explanatory diagram showing an example of a circumscribed rectangle obtained in the labeling in step S36 of FIG. 3;

FIG. 13 is an explanatory diagram showing an example of a high area ratio in step 37 of FIG. 3;

FIG. 14 is an explanatory diagram showing an example of a low area ratio in step 37 of FIG. 3;

[Explanation of symbols]

 11, 21 TV camera 12, 22 Image input device 13, 23 Host computer (personal computer) 14 Moving image extracting means 15 Moving object determining means 17 Monitoring monitor 18 Alarm generating device

 ────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-63-157593 (JP, A) JP-A-62-136991 (JP, A) JP-A 64-65486 (JP, A)

Claims (1)

(57) [Claims] 1. A method for extracting a moving image by obtaining a level difference at a relative position between different frames of an image taken by a television camera, the method including the presence or absence of a moving object based on the extracted moving image the mobile monitoring device for detecting and a latest frame, a moving image of a plurality of frames past the predetermined following the latest frame determined, said kinematic
The level of the latest image for each part of the image
From the level of the corresponding part of the plurality of frames.
Subtract the maximum value of the
It makes reference a moving image extracting unit, a different predetermined size by an image position for extracting a moving image was divided take mobile monitoring unnecessary for the periodic motion in the limited range issue
In addition, within the circumscribed rectangle for each group of extracted images
Moving object determining means for judging whether an image obtained by the moving image extracting means corresponds to a moving object to be monitored by referring to a ratio of the moving object to the moving object.