Automatic Camshift tracking algorithm based on fuzzy inference background difference combining with twice searching
Xiao Gang, Yong Hui Chen, Chen Jiu-jun, Fei Gao · 2010
In order to overcome the shortcoming that traditional Camshift needs artificial orientation during tracking, this paper proposes a new approach of Camshift tracking algorithm based on fuzzy inference background difference. In this paper, the object contour extracted by background difference rather than artificial selection, is used as initial search window so as to realize automatic Camshift tracking. Meanwhile, to avoid object divergence and object losing when the object moves too quickly, twice Camshift searching is combined with background difference to enlarge the search window automatically to ensure consistent targeting. Furthermore, this paper also introduces contour marking and multiple Camshift trackers to implement successful multi-object tracking. Methods mentioned above prove themselves efficient and automatic in tracking one or more moving fishes during the experiments.