Effect of Global Visual Features on Scene Segmentation for Videos

J. Zhou and W. Tavanapong (USA)


: Video segmentation, multimedia databases, browsing and retrieval, feature extraction.


Automatic video segmentation is the first and necessary step for organizing a long video file into smaller units for subsequent browsing and retrieval. The smallest basic unit is shot–a contiguous sequence of frames recorded from a single camera operation. Relevant shots are then grouped into a high-level unit called scene that conveys some mean ing to viewers. In this paper, we first give a stricter scene definition. We then investigate the performance of two re cent scene segmentation techniques. Both utilize visual features of entire video frames for grouping shots into scenes. Our experimental results on two full-length films reveal an interesting insight, suggesting a better approach for scene segmentation.

Important Links:

Go Back