Overlapped-vision-field multi-camera cooperative target positioning and tracking system

Abstract

The invention discloses an overlapped-vision-field multi-camera cooperative target positioning and tracking system. Pixel points in a side length range of the upper side of a rectangular frame of a foreground area are sampled, and combined with geometric constraint of multiple cameras to acquire target candidate foot points, meanwhile, peaks of the lower side of the rectangular frame for expressing pedestrians not in a shielded view angle between the targets are projected, and a center of gravity of each projection point is computed, and at last, the candidate points are weighted, and the position of the target is positioned by using the centers of gravity and the weighted candidate foot points. Compared with target tracking by using the single camera, the system using the multiple camerashas the advantages of wide monitoring range and many observation view angles, can acquire a robust tracking effect by using multi-camera environment constraint and multiple foreground information, and well solves the shielding problem appearing in a tracking process; therefore, even the target is in the shielded view angle, the target can still be positioned and tracked highly accurately, and good market application value is provided.

Description

technical field [0001] The patent of the present invention relates to the field of intelligent video surveillance, and in particular to a target positioning and tracking system with overlapping view areas and multi-camera coordination. Background technique [0002] With the rapid development of smart cities and smart transportation, intelligent video surveillance technology is in a stage of rapid development and has made significant contributions to the cause of security monitoring. However, the gradual increase in the number of monitoring cameras and the monitoring range has increased the labor intensity of camera installation, offline calibration, measurement, and video monitoring analysis. In addition, with the complexity and changeability of surveillance scenarios, traditional single-camera object tracking and localization methods face enormous challenges in practical applications. In recent years, the multi-camera collaborative video surveillance technology has attract...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, the target space positioning system based on the homography matrix and the target space positioning system based on the main axis, in the case of severe occlusion, the target cannot be separated independently in the foreground, and a foreground area may contain multiple targets. At this time, if you use a rectangular frame to frame the foreground area, then this rectangular frame may contain multiple targets. At this time, the four vertices of the rectangular frame cannot replace the foreground information of any target, or it is wrong foreground information. Then use this The four vertices of the rectangular box cannot get reliable object localization results

Images