Video news face retrieval based on Web image datasets

Automatic annotation of broadcasting news programs is a challenging task for multimedia press review. Usually in a video stream, the head of a person moves continuously and changes in facial expressions, lighting conditions, and camera motion produce significant distortions in the image appearance that can largely affect recognition performances. In this paper a system for automatic face identification in TV broadcasting programs is proposed. The proposed approach is based on a joint use of Scale Invariant Feature Transform descriptor and Eigenfaces-based approach. The algorithm has been tested on two national broadcasting channels. Experimental results show that the joint use of these two approaches improves the recognition rate.