JP2922904B2 - Monitoring method by TV camera and monitoring device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a monitoring method for detecting when an abnormal moving object enters a monitoring area by two or more TV cameras, And the monitoring device. The above abnormal moving object means that it does not normally exist in a steady state. Therefore, an abnormal moving object is not always an object accompanying a failure.

The abnormal moving object to be monitored by the present invention is, for example, smoking, water leakage, a person, and the like.

[Conventional technology]

Conventional techniques for automatically detecting the coordinates of an object by performing image processing on image information captured by using two television cameras include: (i) a television as disclosed in JP-A-58-70108. The surface of the object is illuminated by a spot located in the vicinity of the object separately from the camera, and while this is scanned automatically or manually along the shape of the object, the reflected light is image-processed by two TV cameras to determine the position of the object. Technology for measuring shapes and shapes.

(Ii) As disclosed in JP-A-60-140113, the object is image-processed by making the optical axes of two TV cameras parallel, and the respective images are superimposed on the basis of the principle of triangulation from the shift amount. (Iii) using two TV cameras, extracting the object by pattern matching, finding the center coordinates, and triangulating the object by triangulation from the directional angle of each TV camera. Techniques for measuring position are known.

[Problems to be solved by the invention]

Among the known techniques (i), (ii) and (iii), in the method using (i) a spot, (i) the method can be used only at a spot where a spot light is applied.
A moving object reflected at an arbitrary position in the field of view cannot be extracted.
Also, it takes a spotlight scanning time to find the center of the object.

(B) If the object has irregularities, measurement cannot be performed because reflected light cannot be obtained.

 (C) A spot light generator is required, resulting in a large size.

(Ii) In the method of processing an object with two TV cameras with the optical axis parallel, when the distance between the TV cameras is up, down, back and forth, left and right with respect to the target, or when the optical axis is not parallel However, the shape, position, size, etc., of the images captured by each TV are different, and the mere superposition of the images is inferior in accuracy, and enormous time is required for accurate measurement.

(Iii) Extracting a moving object from the background of an object outdoors or the like, including the method (ii) above, involves a large amount of disturbance and requires an enormous processing time.

There was a problem such as, and it could be used only under specific conditions.

The present invention has been made in view of the above circumstances, does not require a spot facility, and detects a moving object within the field of view of a TV camera, and instantaneously and accurately detects its coordinates. It is an object of the present invention to provide a monitoring method that can be measured and an apparatus suitable for performing the method.

[Means for solving the problem]

The basic principle of the present invention created to achieve the above object will be briefly described below.

In practicing the present invention, a conventional visible light TV camera or infrared TV camera can be used.

For example, when monitoring the appearance of a moving object such as a person or smoke (an object that does not normally exist in a steady state), a person or smoke can be detected by a visible light camera or an infrared camera.

According to the present invention, information indicating a state when such a person or smoke (this is an example) does not exist is stored in advance.

Then, the information before the appearance of the moving object is stored every moment as a binarized or multi-valued image processing video, and a difference calculation from the previous image is performed.

When the moving object appears in the field of view, a difference image appears due to a difference in light amount or heat of the moving object with respect to the information before the appearance. Therefore, if the output voltage corresponding to the light amount or the heat exceeds a certain level and is higher or lower than a certain level, it is determined as a moving object, and the level is binarized as a threshold value. .

Next, the difference image having the fixed area is reduced uniformly by a fixed pixel in the Χ direction and the Y direction of the screen. Finally, the above-described reduction processing is stopped while leaving only one pixel or a fixed value for each of Χ and Y, and the center of the area at this time is set as the center position of the moving body.

Next, the azimuth from the center of the image to the center of the moving object is automatically calculated.

The above processing is performed on the information of each of the two TV cameras, and the position of the moving object is automatically calculated based on the principle of triangulation.

[Action]

The object reflected on the TV camera or the infrared TV camera is extracted from the background by a difference from the background. Here, the shape of the moving object is a two-dimensional shape having a certain area, and a small amount of light or heat is generated around the periphery due to a luminance or heat level difference generated by a state change due to a temporal change of the background before and after the moving object appears. The difference image remains as noise.

From this state, all the images on the image are reduced uniformly in the horizontal and vertical directions. At this stage, the surrounding minute noise is removed. The remaining moving object image finally becomes a dot-like or linear image of one pixel level. If the reduction process is stopped at a position where one pixel level remains for each of Χ and Y, the moving object becomes a dot. In this process, the image of the moving object is uniformly reduced in the horizontal and vertical directions from the outer peripheral side, so that the coordinates of almost the center of the image area remain, and as a result, the center coordinates of the moving object are calculated. You. Since all of the above are automatically processed based on information stored in the image memory, instantaneous measurement is possible. Therefore, the moving object reflected in the field of view can be reliably extracted and its position can be calculated.

〔Example〕

Next, an embodiment of the present invention will be described with reference to FIG. 1 to FIG.

This example is an example of monitoring a moving object having a higher temperature than a background object using an infrared TV camera. In the present invention, monitoring refers to the work of finding and locating an object that does not exist in principle in a steady state.

FIG. 1 is an explanatory view of the gist of the present invention, and FIG. 2 is an explanatory view of a schematic arrangement in the apparatus of the present embodiment.

As shown in FIG. 2, a remote point 7 which is a blind spot (area where direct viewing is not possible) with an obstacle 8 separated from a moving object is set as a monitoring base. Infrared TV cameras 5 and 6 are respectively installed at two positions (coordinates of the positions are assumed to be known) away from the remote point 7.

The coordinates of the TV camera 5 are (x ₁ , y ₁ , z ₁ ), and the coordinates of the TV camera 5 are (x ₂ , y ₂ , z ₂ ).

In practicing the present invention, it is also possible to install three or more TV cameras.

The image based on the distribution of the infrared radiation intensity captured by the two TV cameras 5 and 6 is in a multi-valued gradation state as indicated by 1 in FIG. 1 (A _-1 ). It cannot be determined whether there is a moving object. Even if it is cut off by a simple threshold, it is not known whether there is a moving object there. Therefore, one feature of the present invention is to determine the appearance of a moving object based on the state before the moving object appears. That is, it is considered that the moving object (for example, a person or smoke) has some heat, and usually has a higher temperature than the background. (As will be described in detail later, the present invention can be applied even at a lower temperature than the background).

When the mobile enters the monitored area (visual field of the TV camera), as one of the threshold peak of the infrared intensities under the background as shown in FIG. 1 (A _-2), at predetermined time intervals Compare with new image. When a moving object appears, the difference from the threshold value is output as the difference between the multi-valued images as indicated by 2 in FIG. 1 (B _-1 ). FIG (B _-2) is a table showing the difference between the threshold 4 "is binarized background noise.

Therefore, when a portion having a positive output, such as 3 shown in FIG. 1 (C _-1 ), is uniformly displayed as a constant voltage, and the other portions are displayed as a binary value, the moving object is completely extracted. Thereafter, the horizontal of the image, readily punctate by vertical reduction operation or the like, the center position n _x of the moving object on the image, a n _y can automatically calculated, also simply all based on the background image The same function is provided by calculating the difference between the pixel and the coordinates before the appearance of the moving object.

Further, according to the difference method, there is an advantage that the moving object is detected as a difference even if the temperature of the moving object is lower than the background.

In this case, the same identification as described above is performed by taking the absolute value of the difference.

4 is a reduced background noise. Fig. 1 (C- ₂ )
_Shows the signal waveform in the above (C _-1 ). w is the signal width (time width) of the reduced image. This (C- ₂ )
The signal shown by the dashed line in the figure represents the signal before reduction.

Next, the camera angle of view is determined from the result of the moving object identification operation described above. The reduced image 4 of the moving object shown in FIG. 1 (C _-1 )
From the center coordinates n _x, n _y is obtained. From this, the horizontal angle θ ₂ = k ₂ · n _{x the} vertical angle θ ₁ = k ₁ · _ny (where k ₁ and k ₂ are constants determined by the camera). So, using this angle of view with two TV cameras,
Seek the tertiary coordinates of the moving object as viewed from a remote location O _0. Two
If the reference positions of the TV camera are O ₁ and O ₂ reference directions in one direction, the angle of view information is used, and the values shown in FIG. θ ₁ and θ ₂ can be calculated.

That is, the horizontal angle and the elevation angle as viewed from O ₀ , or their absolute coordinates can be calculated. Obviously, the speed of the moving object can be calculated by looking at the coordinates of the two points of the moving object at regular time intervals. FIG. 4 shows a specific example of a system for performing the above calculations.

Video information from the infrared TV cameras 5 and 6 is alternately sent to the video input / output device via the communication device. An image digitally processed as a multi-tone level is input to a video memory, and the multi-valued image 1 described with reference to FIG.
Is stored in such a content.

Therefore, the multi-value image is subjected to the calculation of the moving object extraction of the by the preprocessing circuit 15, where theta ₁ and theta ₂ and is calculated. Next, the camera in program memory 17
By 5,6 coordinates and direction of the program on the basis of θ ₁ and θ
_{The two-} dimensional coordinates of the moving object viewed from the point O ₀ are calculated using the _two . All of these series of operations are automatically performed instantaneously from extraction to calculation.

According to the above-described embodiment, the TV camera arrangement is not particularly restricted, and without having to illuminate a spot, a wide area is monitored by two TV camera images, and when a moving object appears, an automatic Can be detected and located.

〔The invention's effect〕

If the monitoring method of the present invention is used by using the monitoring device of the present invention, it is not necessary to provide a spot facility, and a moving object that is in the field of view of the TV camera is detected, and its coordinates are instantaneously located with high accuracy. It has an excellent practical effect.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of an embodiment of a monitoring method according to the present invention, FIG. 2 is a schematic layout diagram of an embodiment of a monitoring device according to the present invention, and FIG. FIG. 4 is a block diagram of the monitoring apparatus according to the embodiment. 1 ... multi-value image, 2 ... multi-value image of moving object, 3 ... binary image of moving object, 4 ... reduced image of moving object, 5,6 ... TV camera,
7 remote point, 8 obstacle, 9 communication device, 10
... Switch, 11 ... TV camera controller, 12 ... Image input / output device, 13 ... Image memory, 14 ... TV monitor, 15 ...
Preprocessing circuit, 16 output device, 17 program memory, 18 processing unit, 19 image processing device, 20 distance sensor.

──────────────────────────────────────────────────続 き Continued on front page (58) Field surveyed (Int.Cl. ⁶ , DB name) H04N 7/18

Claims (7)

(57) [Claims] 1. A monitoring method for capturing an image of a moving object by two or more TV cameras and calculating the position of the object by image processing, comprising: (a) two TV cameras having known spatial coordinates and directions. And the image information by infrared camera or visible light camera,
(B) comparing the input information with the previous information, and constantly performing a difference calculation; (c) a level within a predetermined range, and When difference information of a given area or more is obtained, the difference information is detected as a moving object, (d) the entire screen by the TV camera is reduced, and the moving object image is pointed, (e) A monitoring method using a TV camera, wherein a position of a pointed moving object is calculated by a triangulation method. 2. A monitoring area according to claim 1, wherein an imaging range of said two or more TV cameras is a monitoring area, and a calculating means is provided at a position where said monitoring area is a blind spot separated by an obstacle. A monitoring method using a TV camera, wherein image information of the TV camera is input to the arithmetic unit by a communication unit. 3. The monitoring area from the blind spot position according to claim 1, based on image information captured by the two or more TV cameras and position information on a detected object calculated by a calculating means provided at the blind spot position. A method of monitoring by a TV camera, wherein the same image information as in the case of directly viewing is created by calculation and output to a monitor. 4. The method according to claim 1, wherein the difference calculation in the item (b) is performed by using one of levels of luminance and heat information of an image.
A monitoring method using a TV camera, wherein the monitoring is performed by calculating a difference between a level in a stationary state stored in advance and a level after the detection of the moving object for each pixel. 5. The method according to claim 1, wherein the difference calculation in the item (b) is performed by using any one of a video, luminance, and heat information as a level.
A surveillance method using a TV camera, wherein a difference between a level in a steady state stored in advance for all pixels and a level aligned with a peak point after the detection of the moving object is calculated for all pixels. 6. The method of claim 1, wherein the step (d) is to reduce the image from the outermost periphery of the pixel until an image of one pixel in each of the Χ direction and the Y direction remains on the screen. A surveillance method using a TV camera. 7. A surveillance device that captures an image of a moving object by two or more TV cameras and calculates the position of the object by image processing. (A) Two TV cameras whose spatial coordinates and directions are known And the image information by infrared camera or visible light camera,
(B) an image memory means for storing digital information every moment; (b) an arithmetic unit for comparing the input information with the previous information and performing a difference operation in a steady state; and (c) a level within a predetermined range. And calculating means for detecting the difference information as a moving object when the difference information is equal to or larger than a predetermined area, and reducing the entire screen by the TV camera to convert the moving object image into a dot. And (d) an arithmetic unit for calculating the position of the pointed moving object by a triangulation method.