Adaptive classifier selection on hierarchical context modeling for robust vision systems

Abstract

This paper proposes a hierarchical image context based adaptable classifier ensemble for efficient visual information processing under uneven illumination environments. In the proposed method, classifier ensemble is constructed in two stages: i) it distinguishes the illumination context of input image in terms of hierarchical context modeling and ii) constructs classifier ensemble using the genetic algorithm (GA). It stores its experiences in terms of the illumination context hieratical manner and derives artificial chromosome so that the context knowledge can be accumulated and used for identification purpose. The proposed method operates in two modes: the learning mode and the action mode. It can improve its performance incrementally using GA in the learning mode. Once sufficient context knowledge is accumulated, the method can operate in real-time. The proposed method has been evaluated in the area of face recognition. The superiority of the proposed method has been shown using international face database FERET