Crowd flow analysis method based on depth camera

Abstract

The invention provides a crowd flow analysis method based on a depth camera. The method comprises the following steps of forming a depth map D of multi-element binomial regression-fitting virtual ground; acquiring a target foreground map F; calculating a height map H corresponding to a foreground image; carrying out over-segmentation on a target area of a depth image; aggregating segmentation areas and acquiring a pedestrian detection result; and carrying out target tracking and flow statistics. Compared to an existing method, by using the method of the invention, a lot of data in a database does not need to be trained; a mathematic model is simple and effective; a flow size of a crowd in the area can be accurately analyzed and intuitive performance is high; and a calculation amount is small and the method is suitable for real-time visual sense monitoring. The method can be widely used for monitoring and management of public places with a large crowd flow, such as a bus, a subway, a square and the like.

Description

technical field [0001] The invention relates to pattern recognition, in particular to a crowd flow analysis method based on a depth camera. Background technique [0002] With the rapid development of computer technology and digital electronic technology, visual monitoring technology has been more and more widely used. Traditional visual surveillance systems can only provide video capture and storage functions, requiring monitoring personnel to watch the display in the monitoring room to find abnormalities, which not only increases the cost of monitoring, but also causes visual fatigue caused by long-term viewing of the display. The vigilance of the system is reduced, so that the monitoring system cannot play its due role at some critical moments. At present, intelligent visual surveillance technology is on the rise, and is getting more and more attention from the media. Intelligent visual monitoring technology is to let the computer replace the human brain, let the camera ...

AI Technical Summary

Problems solved by technology

Traditional visual surveillance systems can only provide video capture and storage functions, requiring monitoring personnel to watch the display in the monitoring room to find abnormalities, which not only increases the cost of monitoring, but also causes visual fatigue caused by long-term viewing of the display. The vigilance of the system is reduced, so that the monitoring system cannot play its due role at some critical moments

Images