Tempo-based MTV-Style Home Video Authoring

Automatic authoring of MTV-style home video using visual and music tempo analysis is studied in this work. In the proposed system, the input home video is first segmented into shots by low level features such as the color histogram difference. Then, three types of tempo analysis are conducted; namely, music, global visual (the frame level) and local visual (the facial expression level) tempo analysis. For local visual tempo analysis, we propose a scheme to compute the facial tension from a sequence of facial appearance variation. Finally, the authoring methodology is presented, which consists of music and visual tempo matching to product MTV-style video. Experiments are conducted using baby home video with encouraging results obtained