Human body target tracking method in nonoverlapping vision field multi-camera network

Abstract

The invention discloses a human body target tracking method in a nonoverlapping vision field multi-camera network, which comprises the following steps of target detection of a single camera, target tracking of the single camera, and target association and continuous target tracking of multi cameras, wherein the target association of the multi cameras comprises the following substeps of target characteristic extraction and topological relation establishment, the target characteristic extraction is extraction of each human body movement target HSV (Hue, Saturation and Value) characteristic, andthe topological relation establishment is establishment of the space-time topological relation of the multi-camera network according to the obtained association relation of targets. By the human bodytarget tracking method disclosed by the invention, the continuous tracking of human body movement targets in a large scene is realized, and the problem of the monitoring of the limited vision range of a single camera is solved.

Description

technical field [0001] The invention relates to a human target tracking method in a non-overlapping multi-camera network, which adopts technologies such as image processing and computer vision, and belongs to the technical field of video monitoring. Background technique [0002] As an all-weather, all-round, real-time and intelligent monitoring method, video surveillance has gradually become an important means to assist public security departments in fighting crime, maintaining social stability, and implementing traffic management. The video surveillance system is mainly composed of two parts: 1) Intra-camera tracking, that is, tracking objects within a single camera field of view; 2) Inter-camera tracking, that is, performing tracking of targets between camera fields of view. At present, the research of intra-camera tracking mainly focuses on key issues such as background modeling, foreground detection, shadow removal, and occlusion processing, and has achieved good results...

AI Technical Summary

Problems solved by technology

The above methods aim to find a series of associations of the continuous trajectories of the same target in different camera fields of view in the surveillance network, but ignore the fact that there are some objects that transfer between camera fields of view during movement, and some The target may disappear in the blind area of ​​the monitoring network during the movement, and there may be new targets entering the monitoring network

Therefore, the topological relationship estimated based on the one-to-one matching association information is not completely correct. In addition, when the monitoring environment changes (addition, deletion, movement, etc. of cameras in the monitoring network), the batch training process must be restarted to obtain new topological relationship

Images