JPS62276985A - Picture supervisory system - Google Patents

Abstract

PURPOSE:To identify the size and shape of a variable object by providing plural TV cameras picking up an overlapped area from a different angle from each other and comparing an existing picture with a reference picture at every TV camera. CONSTITUTION:TV cameras 1a, 1b are arranged on nearly same horizontal plane so that the pickup directions are intersected orthogonally. A video signals obtained by the cameras 1a, 1b are digitized by A/D converters 2a, 2b and written in frame memories 3a, 3b. Variable picture element detection sections 5a, 5b compare the picture of reference picture frame buffers 4a, 4b with the picture elements of the memories 3a, 3b at every picture element and decide a picture element having a larger difference from a setting value as a variable picture. The projection of the variable picture element onto X, Y axes is calculated by projection calculation circuit 6a, 6b, the variable object M is extracted as a rectangular area and the result is given to shape deciding circuits 7a, 7b. The circuits 7a, 7b obtain again the projection on the X, Y axes in a rectangular region, the result is handled as the waveform of the projected shape, the similarity of the waveform is evaluated by the distance calculation to decide the state.

Description

Detailed Description of the Invention 3. Detailed Description of the Invention [Technical Field] The present invention relates to an image monitoring system that issues an alarm by detecting a change in a monitoring image by a TV left camera.

[Background technology 1] Conventional image monitoring methods include a method in which a single TV left camera is used to capture an image, and an abnormality is determined when the change in gradation (gradation) of the captured surveillance image is greater than or equal to a set value. there were. This method can be implemented with a simple configuration, but since it simply detects a cluster of pixels with large changes, it cannot accurately capture the position or size of a changing object on an absolute scale. Ta.

Of course, by installing one TV left camera in a high place with a wide field of view, it is possible to know the approximate location of the changing object. In this case, it depends on the relationship between the monitoring site captured by the TV left camera and the TV camera position, and the TV left camera is fixed so that there is a correspondence between the coordinates of the warning area and the position of the changing object on the image for each installation site. , change the coordinates or distance scale for each monitoring site so that you can roughly understand the position of the object that changes from the image! tf2. Set as a region. Then, when a variable object larger than the set size crosses the set warning line, an IP alert is issued.

By the way, in the case of such a method, it is possible to divide the warning area according to multiple warning levels and to track the approximate movement of a changing object, but in order to accurately detect the distance, it is necessary to monitor the TV left camera. It must be installed at a higher location than the site, and therefore there are height restrictions when the monitoring site is indoors. In addition, even outdoors, the TTI range is limited to a narrow range at a relatively low position, and the TV left camera at a high position may show small objects that should be detected, or the TV left camera building support pillar may create a blind spot, or changing objects may be detected from above. By taking pictures,
This method has a disadvantage in that it is not possible to obtain lateral or frontal i[1iff from a low position, which is effective for identification when the changing object is an intruder.

The above method is an image monitoring method using one TV left camera, but a method of capturing three-dimensional images using multiple TV left cameras is to differentiate the image obtained by capturing and
There are methods to create line drawings by converting them into values, and to extract surfaces and identify three-dimensional shapes based on the connections between the intersections of ropes and the correspondence with other camera images, but these methods are fast. It required a computer that could perform parallel processing and had to perform double-current arithmetic processing, so even if it was realized at the research stage, it was not practical.

[Objective of the Invention 1 The present invention has been made in view of the above-mentioned problem α, and its purpose is to provide image monitoring that can easily identify the position, size, or shape of a moving or changing object. There is a method to provide.

[Disclosure of the Invention 1 Invention 1 The present invention includes a plurality of TV left cameras that take images of mutually overlapping depressions from different angles, and detects changing objects by comparing the current image and a reference image for each TV camera. a first means for identifying the position, size or shape of the changing object in the current image with respect to the reference image; and a first means for identifying the appearance state such as the position, size or shape of the changing object on a horizontal plane. and a changing object identifying means for identifying whether the changing objects corresponding to each TV left camera respectively identified by the first means are the same changing object, and the changing object identifying means identifies whether the changing objects are the same changing object. and means for issuing an alarm when the appearance state of the identified changing object identified by the second means meets a predetermined abnormality determination condition.

The present invention will be explained below with reference to Examples.

Figure 1 shows the circuit configuration of an embodiment using the method of the present invention. In this embodiment, two TV cameras 1a, 1
b are arranged on substantially the same horizontal plane so that their imaging directions are perpendicular to each other. Furthermore, if it is necessary to completely separate a plurality of changing objects M, install a third TV camera IC so as to look down on the monitoring site from a high place. In this embodiment, it is determined whether at least one changing object M enters one warning area or not.
It is assumed that the system is monitored using two TV cameras 1a and 1b.

The video signals obtained by the TV cameras 1a and 1b are digitalized by A/D converters 2a and 2b, respectively, and the current images are stored in 7-frame memories 3a and 3b in a format of, for example, 256×25.
Written with 6 pixels and 6-bit gradation. This writing is performed almost simultaneously for each of the TV cameras 1a and 1b, and is updated at regular intervals. The reference image 7 frame crosses 7y4a and 4b are the respective TV cameras Ia.

1b is a buffer for storing the background image or crocodile image that can be taken in order to compare it with the current image, and this stored image is appropriately sampled and stored as the current image 7 frame memory 3g+
3 Updated with the data of b.

Changed pixel detection unit 51.5b is reference image 7 frame patch 7
A4 &, 4b images and current image frame memory 3
a.

3b is compared pixel by pixel, and a pixel with a larger difference than a certain set value Thp is determined to be a changed pixel, and a pixel with a smaller difference than the set value Thp is determined to be a zero pixel, and the changed pixel is set to 1, zero. 1ltli search is (). The projection calculation circuits 6a and 6b each include a change pixel detection unit 5a,
5b'? 1' This is a circuit that calculates the projection of the detected changed pixel (difference image) in the X and Y axis directions and enters and exits the changed object M as a rectangular neck. The shape determination circuits 7a and 7b calculate and correct the projections on the X clay and Y glaze within the extracted rectangular area, treat the shape of the projection as a waveform, and calculate the similarity of the waveforms by distance. Evaluated by total W, etc.
This is to determine the shape. The position detection circuit 8 includes TV cameras la, 1. This is a circuit for detecting the position of a changed pixel in the image taken in Gb, and the projection calculation circuits 6A and Gb are respectively
The changing pixels on the horizontal plane are plotted from the X-shochi projection of .

Shape determination circuit 9 is horizontal! The shape is determined based on the waveform shape of the projection of the changing pixel on the surface, the size of the rectangular M area, and the horizontal ratio of I11, thereby increasing the reliability of tracking changes in the changing object M.

The changing object identification circuit 10 includes shape determination circuits 7a and 7b.
Based on the determination result, it is determined whether the changing object M photographed by the TV camera 1 lit 1 b is the same each time the current image is updated. The motion detection circuit 11 is a circuit for detecting the movement of the changed object M that is determined to be the same object by the changed object identification circuit 10. The abnormality determination circuit 12 outputs the alert level stored in the condition setting unit 13 and a warning! If the appearance state matches the condition by comparing the appearance state such as shape determination and position detection of the changing object M based on the conditions such as the precept area, '? It is designed to issue a report. The abnormality determination conditions include the direction and speed of movement of the changing object M, the appearance position of the changing object M, etc., and a combination of these conditions or a single condition is used as the abnormality determination condition.

Next, the operation of the embodiment circuit shown in FIG. 11 will be explained based on fIS FIGS. 2 to 4.

First, in the embodiment, the data of the changed pixel detected by the changed pixel detection section 58.5b is compressed and encoded by the image compression encoding circuit 15, and transmitted to a distant place through the communication control circuit 14 and the telephone line section. It is decompressed and decoded and monitored.

Figure 2 shows the image taken by the TV camera 1b.

This shows the case of pixels, and in this case, when the changing object M, which is the detected object, moves from the position A in FIG. Assuming that A is the reference image, a changed pixel exists at the position A for the first current image frame. At this time, when the projection calculation circuit 6b calculates the cumulative value of the pixels that change in the Y-axis direction, a projection onto the XMI as shown in the same figure (bl) is obtained, and the cumulative value of the pixels that change in the X-axis direction is obtained. Then, the projection onto the Y glaze as shown in the same figure (C1) is determined.

FIG. 3 shows the case of a change element in an image taken by the TV camera 1a, and in this case, the change object M which is the detected object
'■The position of the changing pixel in the first current image frame and the second current image 7reno, which are facing toward the camera 1a and are temporally different, overlaps, but the changing object M has moved. If the reference image is a background image whose size changes when the detected object does not exist, there will be a changed pixel for the developed image frame of Pt51.At this time, the projection calculation circuit 6a determines the When calculating the cumulative value of pixels that change in the direction, a projection onto the X axis as shown in the same figure (bl) is obtained, and
When calculating the cumulative value of pixels that change in the axial direction, the same figure (cl)
The projection onto the X axis is determined as follows.

It is determined that there is a change if the condition that the width of the section where the size of these projections (the number of pixels changing in the projection direction) is larger than the set value ThQ1 is wider than the set value Ths is satisfied in both the X and Y axes, As the range of the changing object M, all of the changing pixels or the size of the projection is a set value Thα2 (not shown)
Make it a larger section. In this case, the changing object M is each T'V
A rectangle is cut out for each camera 1 at 1 b. This rectangle is expressed by the R-lateral ratio and area, and the shape determination circuits 7a and 7b recalculate the projections on the X-axis and the X-axis within the rectangular area, treat the shape of the projection as a waveform, and calculate the waveform. The shape is determined by evaluating the similarity using distance calculation. By determining the projection of the changing object M of the rock within the rectangular viewing area, it is possible to separate even if there are a plurality of changing objects M at the same time.

Next, in the position detection circuit 8, from the projection on X#I corresponding to the TV cameras 1a and 1b, the X pongee corresponding to the TV camera 1a is set to Xl, and the X axis corresponding to the TV camera 1b is set to
As shown in Figure @4, the changed pixels on the horizontal plane are plotted as shown in A in the same figure.

In this case, since the rotation of the two TV right cameras on the horizontal plane is not determined, change pixels are assigned around the Xl and X2 axes to match the projections on the Xlll1I and X2 axes.

The shape determination circuit 9 determines the shape based on the waveform shape of the projection of the changed pixel on the horizontal plane obtained from the position detection circuit 8, the size of the rectangular area, and the aspect ratio to further improve the reliability of tracking.

The case where the changing object M is in the position (b) of Fig. 2(a) is the same as above, and the projection of the TV camera ill on the X-axis is shown in Fig. 2(b2), and the projection on the X-axis is The projection on the X-axis corresponding to TV right camera a is as shown in Fig. 3 (b2), and the projection on the X-axis is as shown in Fig. 3 (C2), and The changed pixels look like the mouth in Figure 4.

After the shape determination circuit 9 performs shape determination, the change object identification circuit 10 determines whether this changing object M is the same as the previous changing object M. If the object is the same, the motion detection circuit 11 determines whether the changing object M is the same as the previous changing object M. Abnormality judgment circuit 1 detects the moving position and determines whether the position corresponds to the setting conditions for abnormality judgment.
2, and if applicable, a warning will be issued. In the embodiment, this alarm signal is transmitted to a distant place through the communication control circuit 14 and telephone port #i.

Here, depending on the environmental conditions of the monitoring site, TV camera 1a,
If there is an obstacle or shield that blocks the field of view of 1 b and is not detected as a change on the horizontal plane, the changing object M
It may happen that the projection of the object M changes or disappears due to the movement of the changing object M, but since the position is detected by multiple TV cameras 1a and 1b, it is difficult for any TV right camera to take a picture of the changing object M. , can guarantee the lack of other TV right camera detection.

In other words, by registering and setting the data of known obstacles and shields in advance, you can check the position and shape of the changing object on the horizontal plane, in case it is not possible to take a picture with one of the TV right cameras while the changing object M is moving. Even if it cannot be determined, the changing object M can be tracked by estimating the situation. Furthermore, when there are a plurality of changing objects M, even if it cannot be determined that each of them is the same, it is sufficient to set the worst object for the abnormality determination conditions as the object of caution.

Furthermore, if tracking is not performed, it is of course possible to perform abnormality determination at the position detection stage.

[Effects of the Invention 1] The present invention includes a plurality of Tψ left cameras that image mutually overlapping areas from different angles, means for comparing the current image and a reference image for each 'rv camera and detecting a changed object, and a first means for identifying the position, size or shape of the changing object in the current image; and a tjS for identifying the appearance state such as the position, size or shape of the changing object on a horizontal plane.
2 and each T identified by the first means.
A changing object identification means for identifying whether changing objects corresponding to the V right camera are the same changing object, and a tIS of changing objects identified as the same changing object by the changing object identifying means.
Since the present invention is equipped with a means for issuing an alarm when the appearance state identified by means 2 corresponds to a predetermined abnormality judgment condition, it is possible to easily identify the position and shape of a changing object on a horizontal plane. In particular, since multiple TV cameras are used, even if any TV camera other than the changing object is disturbed, the position of the changing object can be easily determined by imaging the remaining TV cameras, and the monitoring area can be limited or , it has the effect of easily removing disturbances, and since the above configuration allows detection of changing objects by projection calculation, there is no need for complex calculations, and complex judgment conditions can be set with one TV camera. This is a practical system that has higher M orderliness than the case where a three-dimensional image is created by processing a plurality of image data. This has the effect of making it possible to build fII.

[Brief explanation of the drawing]

FIG. 1 is a circuit configuration diagram of an embodiment of the present invention, and FIGS. 2 to 4 are explanatory diagrams of the same operation, in which 1 a and 1 b are TV cameras, 5 at S b is a changed pixel detection section, and 6 a,
6 b is a projection calculation circuit, 7 at T b is a shape determination circuit, 8 is a position detection circuit, 9 is a shape determination circuit, 10 is a changing object identification circuit, 11 is an S detection circuit, 12 is an abnormality determination circuit, and 13 is a This is a condition setting section. Agent Patent Attorney Ishi 1) Ai 7 Figure 2 Figure 3 1 x 1 Figure 4 - Kari

Claims (6)

[Claims] (1) A plurality of TV cameras that image mutually overlapping areas from different angles, a means for comparing the current image and a reference image for each TV camera to detect a changing object, and a means for detecting a changing object by comparing the current image and a reference image for each TV camera, and a first means for identifying the position, size or shape of the changing object; a position of the changing object on a horizontal plane;
A second means for identifying appearance states such as size or shape, and changing object identification for identifying whether or not the changing objects corresponding to each TV camera respectively identified by the first means are the same changing object. and means for issuing an alarm when the appearance state of the changing object identified by the second means of the changing object identified by the changing object identifying means as the same changing object meets a predetermined abnormality determination condition. An image monitoring method characterized by: (2) In the above-mentioned changing object identification means, each T
Claim 1, characterized in that it is identified whether the changing objects corresponding to the V camera are the same changing object.
Image monitoring method described in section. (3) A patent claim characterized in that the above-mentioned abnormality determination conditions include conditions such as the direction of movement, speed, and appearance position of the changing object, and abnormality determination is performed based on a combination of these conditions or a single condition. The image monitoring method described in item 1. (4) The first means is to project the difference image between the current image and the reference image for each TV camera in the X-axis direction and the Y-axis direction to determine the position and size of the changing object corresponding to each TV camera. , the image monitoring method according to claim 1, wherein the image monitoring method determines the shape of the image. (5) The image monitoring system according to claim 1, characterized in that two TV cameras are used and these TV cameras are installed so that the shooting directions of these TV cameras are perpendicular to each other. (6) Register and set the presence of known shields, obstacles, etc. that are not detected as changes on the horizontal plane. Changes may occur if there is a TV camera that cannot detect changes while the object is moving, or when there are changes on the horizontal plane. 2. The image monitoring system according to claim 1, wherein when the position and shape of the object cannot be determined, the situation is estimated from the registered and set data and the changing object is tracked.