点击次数：

论文类型：期刊论文

发表时间：2013-01-01

发表刊物：ICIC Express Letters

收录刊物：EI、Scopus

卷号：7

期号：3 B

页面范围：1107-1113

ISSN号：1881803X

摘要：In the visual tracking system, the 'drift' phenomenon always occurs due to the disturbance of the noise, especially in the case of occlusion. To overcome this problem, a novel particle filter algorithm based on the projection histograms and centroid shift (PHCS) is proposed. The algorithm is under the particle filter framework. The projection histograms are utilized to represent the object robustly from the spatio-colorimetric domain, which could discriminate the object precisely from the interference of the nonobject. And the centroid shift scheme is adopted to optimize the particle distribution, which moves the particles toward the optimal importance distribution so as to solve the degeneration disadvantage of the particle filter. With the above two schemes, the observation likelihood function and the particle distribution of the particle filter algorithm are optimized and suitable for realizing the visual tracking task under complex noise. Experimental results demonstrate the robustness of the proposed method in various scenarios. ? 2013 ICIC International.