Method and apparatus for detecting the number of persons

• US 5,121,201 A
• Filed: 08/14/1990
• Issued: 06/09/1992
• Est. Priority Date: 08/25/1989
• Status: Expired due to Fees

• Current Assignee: Fujitec Company Limited
• Original Assignee: Daido Denki Kogyo Kabushiki Kaisha
• Inventors: Seki, Hiroshi
• Primary Examiner(s): Britton, Howard W.
• Assistant Examiner(s): VU, KIM Y
• Application Number:
  US07/567,051
• Time in Patent Office: 665 Days
• Field of Search: 358/107, 358/108, 358/105, 382/22
• US Class Current: 348/143
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 7/66 of image moments or centre of gravity
• CPC Class Codes: G06V 40/10 Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
• CPC Class Codes: G07C 9/00 Individual registration on entry or exit

Abstract

A method and an apparatus for detecting the number of persons in which the differential pictures are converted into binary signals being affected neither by the color of the hair of the persons to be detected nor by the change in the intensity of illumination even when the contrast is small over the background floor. The shape of the picture is corrected, and the area of the picture becomes nearly uniform irrespective of the positions of the persons in the visual field of the TV camera. Moreover, a correlation value relative to the pattern of a predetermined size and a center of gravity of the picture are found while compressing the number of pixels, making it possible to detect the number of persons within a very short period of time maintaining high precision.

Claims

1. An apparatus for detecting the number of persons by processing the picture signals from a TV camera, comprising:

a differential picture operation unit for converting an absolute value of a difference between two consecutive pictures from said TV camera into a binary value;

a shape correcting unit for correcting an area of the picture depending upon a distance between a center of said TV camera and a picture of a person to be detected;

a pixel number compression unit for decreasing the number of pixels by combining a plurality of pixels into one;

a correlation value detecting unit for scanning a pattern of a predetermined size with the pixel of picture produced from said pixel number compression unit as a unit to thereby find a correlation value between said pattern and said picture that is produced; and

a center-of-gravity detecting unit for detecting the center of gravity of a partial picture in said pattern, wherein said partial picture is obtained by removing the picture included in said pattern that has a peak of said correlation value as a center.

2. A method of detecting the number of persons comprising the steps of:

(1) receiving the pictures consecutively from a TV camera maintaining a predetermined time interval, finding an absolute value of a difference between two consecutive pictures, and converting a differential picture thereof into a binary picture with a threshold value, wherein said threshold value is capable of removing background noise to take out only a portion of the picture that has changed with the lapse of time,

(2) correcting the shape of a picture;

(3) combining a plurality of pixels into one, thereby decreasing the number of pixels and shorting the processing time for detecting the number of persons,

(4) scanning the pattern of the size of a person on the picture, wherein the number of pixels has been compressed on said picture to thereby find a correlation value between the pattern and the picture,

(5) detecting the number of persons relying upon peaks in the correlation values between the patterns and the pictures, scanning on the pictures the same patterns as those of when the correlation values were found based on the remaining partial pictures to thereby find centers of gravity of the pictures in the patterns, finding those centers of gravity that are nearly in agreement with the centers of the patterns, counting the centers of gravities while suitably weighting the counted centers of gravities depending upon the distances relative to the positions of persons already detected, and removing pictures included in the patterns having counted centers of gravities as centers, and

(6) adding up together the correlation values between the persons and the patterns and the number of persons counted based on the centers of gravities.