Video Searching System Based on Human Face Identification and Facial Expression Recognition Using Msm and Aam

These days, due to the technological progress being made in digital video devices, demand has been increasing for effective ways to search specific images in large-volume, accumulated recordings. To solve this problem, we propose an efficient video searching system based on human face images. The user will be able to retrieve desired scenes containing the queried person with the specified facial expression in a video file. In the tested system, an actor of a drama is identified and the actor’s facial expression is recognized using the Mutual Subspace Method (MSM) and Active Appearance Models (AAMs). Experiments using drama have yielded results in which the identification rate was 76.8% and the recognition rate of the facial expressions was 84.8%. TOMOKO OKADA, TETSUYA TAKIGUCHI and YASUO ARIKI 42