Application of Improved Bhattacharyya Coefficient Based Multi-Object Detection and Tracking Integrated Strategy
Mingbin Zeng, Zichao Wei, Hu He, Xu Yang · 2018
Moving object detection and tracking has attracted the attention of many researchers. In this study, we proposed an integrated detection and tracking strategy, which used the output of multi-object detection as the candidate locations for the tracking algorithm, thus to avoid repeated detection, redundant detection, and also greatly reduce the searching area for tracking algorithm. We improved the Bhattacharyya coefficient as a similarity function, and improved the traditional Meanshift strategy. Results show that the proposed tracking algorithm can adapt to the two situations of “target crossing” and “object occlusion”, and has a certain degree of robustness.