An Improved Tracking Algorithm for Occlusion Problem Based on STAPLE

摘要

Occlusion is one of the common problems in target tracking, which presents challenges for real-time and robust tracking. In order to solve the problem that the target is lost after being obscured, a STAPLE algorithm combined with SVM is presented in this paper. On this basis, occlusion detection, LBP-based deformation detection and multi-peak repositioning algorithm are added to solve the problem of template contaminated caused by target occlusion and insufficient robustness to distortion. When the target is distorted, the target model is continuously updated to maintain its robustness. When the decrease in confidence level is not caused by deformation, the target detection mechanism is activated and the update of the target model is stopped. The experimental results show that the proposed method is much better than previous algorithms.