Method of monitoring pedestrians and counting based on overhead camera

Abstract

The invention discloses a method of monitoring pedestrians and counting based on an overhead camera in the fields of computer version and pattern recognition. The method comprises the steps that firstly, a person and an object which are close to each other are taken as a tracking object; and secondly, the number of the pedestrians in the region is estimated through analyzing the size and the contour feature of the object region so as to realize the aim of counting. Different from the current known method, the method is characterized in that a plurality of pedestrians which are closer to one another are taken as a tracking object, and the number of the pedestrians in the object is estimated according to the area and the contour feature of the object. At the same time, a method which is based on image contours and is fused with partial gait features is used for differentiating the pedestrians and non-pedestrians (such as luggage) in a foreground block. Proved by experiments, the method provided by the invention has a satisfied counting result under the conditions that the gap among the pedestrians is smaller and non-pedestrian moving objects exist.

Description

technical field [0001] The invention belongs to the field of computer vision and pattern recognition, and in particular relates to a new monitoring and counting method based on an overhead camera. Background technique [0002] In the field of computer vision, the number of applications for video surveillance counting is increasing. In terms of security, it is often necessary to calculate the number of people entering and leaving the building; in terms of intelligent transportation, it is necessary to count the traffic at each intersection to rationally plan traffic facilities; at subway transfer stations, it is necessary to count the number of people transferring in each direction in order to arrange the number of subway trains reasonably . Manual counting costs are high and it is difficult to count accurately when there is a large flow of people, so video surveillance has been applied to the counting problem. [0003] There are already many monitoring and counting systems...

AI Technical Summary

Problems solved by technology

Detecting the features of the head is a feasible method, but detecting the head from the foreground requires a slightly larger amount of calculation, which will affect the real-time performance of the system

Estimating the number of pedestrians contained in it according to the size of the foreground can also be used for monitoring and counting, but due to the fatness and thinness of pedestrians, and other items carried, such as packages, suitcases, children, etc., the size of the foreground area has a great impact , it is easy to cause misjudgment

Images