Adaptive Multi-feature Combination Object Tracking Based on Blocking

ZHOU Fang-yan,TANG Jian,HE Jin-song
DOI: https://doi.org/10.3969/j.issn.1000-3428.2013.04.055
2013-01-01
Abstract:(Abstract)Tracking algorithm based on RGB histogram is sensitive to the similarity color in background, leading to failure in tracking. In view of the above problem, a tracking algorithm based on blocking with multifeature comb ination is proposed. This algorithm adopts KS similarity measure in Edge Ori entation Histogram(EOH) and RGB histogram, with adaptive combination to track object. Experimental results show that it has better performance than single feature as well as robust to occlusion.
What problem does this paper attempt to address?