Automatic face recognition from video sequences using a template based cross correlation method

Face recognition in videos has been an active topic in the field of object recognition and computer vision. In this paper we propose an automatic face recognition algorithm from video sequences using a template based cross correlation (TBCC) method. It utilizes random selection of frames to form the training template for the discriminant feature representation of a face. The proposed method was tested by randomly selecting a subject from the RML and Cohn-Kanade (CK) databases with non-overlapping sequences between test and training sequences. Form the experiments on several datasets, the system performance achieved an average recognition rate of 88%.