JPH01121978A - Objects number detector - Google Patents

Abstract

PURPOSE:To prevent the number of objects from being superposingly detected and to accurately detect it by detecting the objects with plural television cameras whose visual fields overlap each other and subtracting and correcting the number in an overlapping part from the number detected by all the television cameras. CONSTITUTION:Video signals from television cameras 6a and 6b are digitally converted by picture input parts 1a and 1b and object extracting parts 2a and 2b detects objects 8 as projections 9a and 9b. Coordinates calculating parts 3a and 3b calculate coordinates Pa and Pb closet to visual field center coordinates 10a and 10b of the respective cameras 6a and 6b out of the projections 9a and 9b and outputs them to a coincidence number calculating part 4. The calculating part 4 detects whether or not the coordinates Pa and Pb on a digital picture coincide, calculate only the number of coincident coordinates and outputs a calculation result to a correcting part 5. The correcting part 5 subtracts the number from the calculating part 4 from the detected sum of the objects 8, the number of the objects 8 repeated in the overlapping part is corrected, corrected to the original number and the number of objects 8 is detected.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a device for detecting the number of objects, and in particular, a device for detecting the number of objects by installing a plurality of television cameras (hereinafter referred to as “television cameras”) to detect the number of objects in an elevator hall or inside an elevator car. The present invention relates to a device for detecting the number of objects suitable for detecting the number of people in a public place.

[Conventional technology]

A widely used method is to convert a video signal from a television camera into a digital signal and recognize a detection target image from the obtained digital signal image. Japanese Utility Model Application Publication No. 53-56676 discloses a technique for detecting the number of people in an elevator hall or in a car using such a method.

According to this method, an image of the elevator hall or the inside of the car is taken with a single television camera, and the obtained video signal is processed. Therefore, when the monitoring area is relatively small,
This technology could be used as is if the ceiling was high and a single TV camera would suffice.

(Problem to be solved by the invention) However, when monitoring using a TV camera, if the ceiling is low and the TV camera must be installed at a low position, or if the monitoring area is large, multiple TV cameras may be required. It may happen that you have to install a. In such situations, TV cameras should be installed so that their fields of view partially overlap to avoid blind spots.

In the conventional people detection device described above, the number of people can be detected by processing the video signal of one TV camera, but if the devices are placed so that their field of view partially overlaps, the overlapping portions may be overlapped. There was a problem in that it was not possible to accurately detect the number of people in the entire monitoring area because the number of people in the monitoring area was detected.

The present invention was made to solve the above problems, and it is possible to detect the number of people in the entire monitoring area with high accuracy even when multiple TV cameras are installed so that their field of view partially overlaps. The purpose is to obtain a number of people detection device.

(Means for Solving the Problems) The object extraction device according to the present invention includes a coordinate calculation unit that calculates the coordinates of an object image detected from the image of the television camera closest to the field of view of the television camera, and a coordinate calculation unit for each of the television cameras. A matching number calculating part that calculates the number of matching coordinates by comparing the coordinates obtained by the coordinate calculating part, and a matching number calculating part that calculates the number of matching coordinates by comparing the coordinates obtained by the coordinate calculating part. and a correction section that corrects the number of objects.

[Effect]

According to the present invention, even if an object is detected by a plurality of television cameras whose fields of view overlap with each other, the number of objects in the overlapping portion is corrected by subtracting from the number detected by all the television cameras to accurately detect the number of objects.

〔Example〕

The present invention will be explained below based on the embodiments shown in FIGS. 1 and 2. FIG. 1 is a diagram showing the configuration of the apparatus of this embodiment. In the figure, (3) ('la) and (1b) are image input units that manually input video signals from a television camera and output digital signals; (2a); ), (2b) is an object extraction unit that extracts an object (for example, a person), Ha), (3b) is a coordinate calculation unit that calculates the coordinates of the extracted object closest to the coordinates of the center of the field of view of the television camera, (4) is the coordinate calculation unit (3a)
, a matching number calculation unit that compares the coordinate values calculated in (3b) and calculates the number of matching objects, (5)
is a correction unit that subtracts the number of matches calculated by the number of matches calculation unit (4) from the total number of detected objects.

A method for detecting an object using the embodiment apparatus configured as described above will be explained based on FIG. 2. TV cameras (6a) and (6b) are in the X-axis direction (left and right direction of the screen)
TV cameras (6a), (6
As shown in the figure, the visual fields (7a) and (7b) of b) partially overlap to prevent a blind spot from occurring.

The object (
8), a projection (9a) of the object (8) on the floor.

(9b) is formed.

The image signals from the television cameras (6a) and (8b) are converted into digital signals in the image processing units (la) and (lb), and the image signals are converted into digital signals by the image processing units (la) and (lb), and the image signals are converted into digital signals by the image processing units (la) and (lb).
), the object extraction unit (2a), (2b)
project only the object (8) (9a) and (9
b) The detection signal is sent to the coordinate calculation unit (3a
), Hb). Coordinate calculation unit (3a),
(3b) Throw! 2 (9a), (!lb), the coordinates (P
a) Calculate , (Pb) and output the result to the matching number calculating section (4). Here the coordinates (Pa).

(pb) represents the same point indicating the coordinate (P) on the floor surface of the object (8). Here is the TV camera (6a).
By setting the positional relationship of (8b) with respect to the floor surface in advance, the positions on the digital image corresponding to the projections (9a) and (9b) of coordinates (P) are automatically determined. Therefore, based on the input from the coordinate calculation units (3a) and (3b), the match number calculation unit (4) calculates the coordinates (Pa) on the digital image.
), (Pb) are matched, only the number of matched coordinates is calculated, and the calculation result is output to the correction unit (5). The correction unit (5) subtracts the value from the matching number calculation unit (4) from the sum total of objects (8) detected by each television camera (6a), (a, b), and calculates the duplicate objects in the overlapping area. The number of objects (8) is corrected to the original number to calculate the number of objects (8).

As described above, according to this embodiment, the number of objects detected duplicatedly among the objects detected by a plurality of television cameras is corrected.
Accurate number of detections can be obtained.

〔Effect of the invention〕

As described above, according to the present invention, the number of objects can be detected with high accuracy since objects are imaged by a plurality of television cameras and objects in the overlapping field of view of each camera are not detected redundantly.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the target wheel number detection device according to the present invention, and FIG. N2 is a diagram for explaining the detection state. In the figure, (3a) and (:+b) are coordinate calculation units,
(4) is a matching number calculation unit, (5) is a correction unit, (6a),
(6b) is a TV camera, (8) is an object, (9a)
, (9b) is the digital image of the projection, (10a)
, (10b) are visual field center coordinates, (Pa), (Pb
) are the coordinates closest to the visual field center coordinates.

Claims (1)

[Claims] The video signals of multiple television cameras arranged so that their field of view partially overlaps are each converted into digital signals, and an object image is extracted from the obtained digital signal image and captured within the entire field of view of the television camera. In the object number detection device for detecting the number of objects, a coordinate calculation unit calculates the coordinates closest to the visual field center coordinates of the object image detected from the images of each of the above-mentioned television cameras, and a coordinate calculation unit calculates the coordinates determined by each coordinate calculation unit. a matching number calculation unit that calculates the number of matched coordinates by comparing the numbers, and a correction unit that corrects the number of objects by subtracting the numerical value of the matching number calculation unit from the total number of objects detected by each of the television cameras. A target object number detection device comprising: