Real-time robust tracking with part-based and spatio-temporal context

Abstract

Owing to the significant and excellent performance of correlation filter in the aspect of computation convenience, correlation filters based trackers have become increasingly popular in the visual object tracking community. However, complete or partial occlusion is one of the major factors that seriously impact the tracking performance in visual tracking. To address this issue, we propose a novel tracking algorithm that perfectly integrates the results from the global correlation and local correlation filters for estimating the more accurate position of target. Then, we introduce the occlusion detection mechanism to eliminate the occlusion impact on the final position of object. In addition, our proposed tracker employs the spatial geometric constraints among the global object and local patches of object for preserving the structure integration of object. For verifying our method, we conduct extensive qualitative and quantitative experiments on challenging benchmark image sequences.