Title: Research and analysis of video image target tracking algorithm based on significance

Authors: Heshuai Shao; Jie Shen; Zhen Zhang; Huiyi Liu

Addresses: College of Computer and Information, Hohai University, Nanjing, China; College of Computer and Software Engineering, Huaiyin Institute of Technology, Huaian, China ' College of Computer and Information, Hohai University, Nanjing, China ' College of Computer and Information, Hohai University, Nanjing, China ' College of Computer and Information, Hohai University, Nanjing, China

Abstract: A target tracking algorithm based on joint probabilistic data association encounters the problems of target loss and combinatorial explosion in cluttered and target-intensive tracking environments. In order to address these challenges, this paper proposes a video target tracking algorithm based on significance joint probabilistic data association. This algorithm detects the moving target in a video sequence through significance calculation. The detection results are classified, and the returned classes are then used as valid echoes for data association. Meanwhile, the validation matrix and joint probability in the joint probabilistic data association algorithm are redefined based on the significance information and a validation matrix and association probability based on significance is proposed. The association results are compensated and corrected using the colour association probability. Experiments in real scenes demonstrate that the proposed algorithm can suppress clutter in the background, simplify interconnected events, and solve the problem of computational combinatorial explosion.

Keywords: target tracking; data association; significance; validation matrix; association probability; colour association probability.

DOI: 10.1504/IJHPSA.2018.094149

International Journal of High Performance Systems Architecture, 2018 Vol.8 No.1/2, pp.82 - 93

Received: 23 Nov 2017
Accepted: 16 Apr 2018
Published online: 17 Aug 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article