Clear Face Analysis from MPEG Compressed Video

Ling-Yu Duan,Qi Tian
DOI: https://doi.org/10.1145/641007.641028
2002-01-01
Abstract:In this demonstration, we present a system to analyze the clear degree of faces present in MPEG compressed video of Head-and-Shoulders style. The proposed system consists of three hierarchical modules: low-level features extraction, robust face tracking, and clear faces selection. We have integrated the core algorithm into an Automated Transaction Service (ATS) surveillance system. The Incremental Focus of Attention (IFA) architecture is taken to combine pixel domain processing with compressed domain processing --- thus, implemented system exhibits computational efficiency and tolerance to very cluttered scenes. The proposed system has successively detected segments with clear frontal faces from more than 20 Automated Teller Machine (ATM) testing clips in MPEG format, each of which consists of 1~3 transactions. In addition, the proposed scheme implies some potential video mining applications, such as automatic checking to verify entry authorization, retrieval of suspicious activities in prerecorded video surveillance sequences.
What problem does this paper attempt to address?