Abstract: Video-based face recognition and tracking are the research hotspots of computer vision. 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 image in video, large variation of face scale, is the focus of this paper. Firstly, a face recognition algorithm using the Associative Neural Network is introduced. And then, the algorithm is merged into TLD (Tracking-Learning-Detection) to improve the performance. The experimental results demonstrate that the improved algorithm is successful and practical.


Key words: ace recognition, face tracking, associative neural network, TLD