Suspicious person cross-lens tracking method and device based on spatial constraint

Abstract

The embodiment of the invention relates to the technical field of video monitoring, and particularly discloses a suspicious person cross-lens tracking method and a device based on spatial constraint.According to the suspicious person cross-lens tracking method based on spatial constraint, pedestrian targets in the video are detected in real time by adopting a deep learning algorithm; a selected suspicious personnel target concerned by the user is selected by utilizing the interactive operation; according to the method, single monitoring camera tracking of the selected suspicious person and cross-lens tracking taking pedestrian features extracted based on a deep learning algorithm as a comparison basis are realized, global optimization is realized in a local scene according to the positioninformation and the motion information of the selected suspicious person target, and the pedestrian retrieval and re-identification efficiency is effectively improved.

Description

technical field [0001] The invention relates to the field of video surveillance, in particular to a method and device for tracking suspicious personnel across shots based on space constraints. Background technique [0002] The existing pedestrian re-identification technology is mainly based on the basis of multi-target tracking to obtain the tracking trajectories of multiple cameras and multiple targets, and then use the deep learning algorithm to extract the average characteristics of pedestrian trajectories, and finally achieve multi-camera pedestrian detection by aggregation of pedestrian average features. The pairing of trajectories finds out the trajectories of the same pedestrian and connects them. [0003] In the field of video surveillance, more attention should be paid to the trajectories of a single or a few special suspects in the shot. Cross-camera tracking of all pedestrians in the video is difficult in system implementation and requires a lot of computing perfo...

AI Technical Summary

Problems solved by technology

[0003] In the field of video surveillance, more attention should be paid to the trajectories of a single or a few special suspects in the shot. Cross-camera tracking of all pedestrians in the video is difficult in system implementation and requires a lot of computing power

At the same time, the existing pedestrian re-identification technology is more inclined to process the tracking trajectory in the later stage, lacking the visual sense of tracking

Images