Real-time classification in tracking human using segmental feature and particle filter

This paper propose the efficient method concerning verifying and tracking of human face. Recently, there has been much interest in automatically face recognition and tracking in many areas such as intelligent robotics, military, smart device applications and automatic surveillance system. Though there have been many demands about real-time face verification and tracking at the same time, however it is insufficient to research algorithm which accomplish tracking and verification simultaneously. Our goal is to solve these two problems at the same time to save computation time and elevate the performance. This algorithm is consisted of segmental feature and particle filter. It is theoretically based on discriminative common vector method and Fisher's LDA. The algorithm trains segmented and shift face image to obtain new segmental features, then we take orthogonal projection matrix using Gram-Schmidt orthogonal-ization. To solve tracking problem, particle filter algorithm is used.