Video-based face recognition using tensor and clustering

Video-based face recognition using tensor and clustering

JidongZhao1, WanjieZhang2, JingjingLi1,KeLu1


1University of Electronic Science and Technology of China, Chengdu, 610731, China

2Beijing Up-tech Harmony Co., LTD, Beijing, China

Video-based face recognition has become one hot topic in the field of pattern recognition recently. How to fully utilize the spatial and temporal information in video to overcome the difficulties existing in the video-based face recognition, such as low resolution of face images in video, large variations of face scale, radical changes of illumination and pose as well as occasionally occlusion of different parts of faces, is the focus. In this paper, we propose a novel manifold-based face recognition algorithm using tensor and clustering(TCVLPP), which can discover more space-time semantic information hidden in video face sequence, simultaneously make the best of the intrinsic nonlinear structure information to extract discriminative manifold features. We also compare our approach with other algorithms on our own video databases. The experimental results show that TCVLPP can get a higher recognition accuracy rate for video-based face recognition.