Video segmentation with the assistance of audio content analysis

Abstract

Video structure extraction is essential to automatic content-based organization, retrieval and browsing of video. However, while many robust shot segmentation algorithms have been developed, it is still difficult to extract scene structures or group shots into scenes. We present a novel audio assisted video segmentation scheme, in which audio and color information is integrated in video scene extraction. A novel audio segmentation scheme is developed to segment audio tracks into speech, music, environmental sound and silence segments. A robust algorithm for shot grouping based on correlation analysis is also developed to further enhance the scene extraction accuracy.