This paper proposes a human face indexing system for video data with face extraction, recognition, tracking and modeling functions. The proposed system can extract, recognize and track the human faces regardless of their orientation and sizes. The time sections including tracked faces are recorded with their names as indices. These indices can be used to classify or retrieve the video data. If the face could not be recognized the tracked facial images can be used for face modeling. We used subspace methods for the face extraction, recognition and tracking.
展开▼