OAM-RC

Abstract

To improve the accuracy and stability of the moving target tracking in optics target tracking testing system, this paper proposes a particle filter tracking algorithm by using multiple features similarity function information fusion method. Under the basic framework of particle filter, the weighted histogram of target color and edge are applied to describe target features, the edge feature was extracted by Sobel algorithm, then utilizes the edge feature to establish the target’s edge observation model, analyzes the features correlation that is used to measure between different features prediction and observation states. Through the experiment and analysis, compared with the traditional particle filter algorithm, the optimized particle filter algorithm reduces the average error by 26.2 pixels, decreases the standard deviation by 24.7 pixels, the results show that the proposed particle filter target tracking algorithm not only can accurately track the moving target when the target and the background color are similar, but also can stably tracking target when the target rotates and the target is occludes partially, which can verify the method of the particle filter tracking algorithm based on the multiple feature similarity function information fusion is advanced..

Keywords

Particle filter, Target tracking, Information fusion, Similarity function.

Citation

HANSHAN LI, Particle filter target tracking algorithm based on multiple features similarity function information fusion method, Optoelectronics and Advanced Materials - Rapid Communications, 13, 11-12, November-December 2019, pp.598-605 (2019).

Submitted at: Dec. 14, 2018

Accepted at: Dec. 10, 2019