Automatic Tracking of Face Sequences in MPEG Video

Human faces are commonly found in video streams and provide useful information for video content anal-ysis. This paper presents a robust face tracking system to extract multiple face sequences from MPEG video without human intervention. Specifically, a view-based DCT-domain face detection algorithm is first applied periodically to capture mostly frontal and slight slant-ing faces of variable sizes and locations. The face tracker then searches the target faces in local areas across frames in both the forward and backward direc-tions. The tracking combines color histogram match-ing and skin-color adaptation to provide robust track-ing. This paper focuses on developing effective tech-niques that exploits the features in DCT domain and the characteristics of video compression standards like MPEGs. The effectiveness of the algorithm is demon-strated using a range of videos obtained from multiple sources like the news and movies.