Robust face tracking by motion energy and genetic algorithms

This paper presents a robust approach to the localization and tracking of a human face in image sequences. First, a new motion energy representation is defined and used to extract the human region in real time. Then the hair and the head knowledge are used to locate the face area. For face tracking, a new genetic algorithm-based dynamic template-matching method is applied to search the new position of the face in each new video sequence. Experimental results show that the proposed method is robust to illumination, face poses, facial expressions and image distractors such as facial occlusion by hands.