A spatio-temporal context target tracking method based on levy points

Abstract

The invention discloses a spatio-temporal context target tracking method based on feature point assistance, comprising the following steps: S1, selecting a target frame in an initial frame, and initializing a spatio-temporal context model and a target feature point model S2, using feature point matching in the next frame Track the target feature points with the optical flow tracking method, and obtain the target feature point set through clustering, and obtain the target estimated location area; S3, establish a local context appearance model in the target estimated location area, and then calculate the correlation with the spatio-temporal context model Obtain the confidence map, and obtain the final position of the target at the maximum value of the confidence map; S4, according to the tracking results, combined with the change rate of the target feature points to determine the degree of target occlusion, and control the update of the spatiotemporal context model. The present invention still has a stable tracking effect and better robustness under the conditions of background interference, occlusion, target rotation and fast movement.

Description

technical field [0001] The invention belongs to the field of computer vision, mainly relates to vision-based target tracking, and specifically relates to a feature-point-assisted spatio-temporal context target tracking method. Background technique [0002] Online target tracking has a wide range of applications in military reconnaissance, video surveillance, behavior recognition, human-computer interaction, and mobile robot tracking and navigation. However, due to the interference of many factors in the tracking scene, such as target rotation, background light changes, fast target movement and occlusion, these factors bring great challenges to dynamic target tracking. Since the target may move out of the field of view, when the target is occluded or lost and reappears, the tracking algorithm needs to be able to detect and track the target again. Therefore, it is of great significance to develop a robust and efficient tracking method. Contents of the invention [0003] Th...

AI Technical Summary

Problems solved by technology

However, due to the interference of many factors in the tracking scene, such as target rotation, background light changes, fast target movement and occlusion, these factors bring great challenges to dynamic target tracking.

Images