Salient region detection based on Adaboost and integration of multi-features space
Yi Zhang, Qichang Duan, Sile Li, Yunxian Ran · 2017
Based on the existing methods, this paper tries to propose a salient region detection based on integrated theory of features and Adaboost algorithm. The integration theory of features indicates that the salient region corresponds to multi-features space, such as color, direction, shape, texture and so on. And the visual system deals with independent features in parallel. The visual system make individual characteristics become a salient area. Adaboost algorithm is able to integrate multiple independent weakly classifiers into a high-performance and powerful classifier. In order to obtain better salient region result, this paper introduces the Adaboost theory to integrate multi-features space.