Audio Classification for Scene Change Detection in Video Sequences

The organization of video databases according to the semantic content of data is a key point in multimedia technologies. In fact, this would allow algorithms such as indexing and retrieval to work more efficiently. The segmentation of a video sequence into scenes and the characterisation of each scene has been suggested as a technique for organising video information. Typically this is performed by analyzing the video signal only. Human beings, on the other hand, use both their visual and auditory systems to perceive the semantics of a film. The associated audio signal can thus be useful to extract information which cannot be simply derived from the sole analysis of the video signal or at least to make the video processing more robust. In this work a technique which uses audio together with video information is proposed in order to improve the performance of existing scene change detection algorithms. Tests are performed on material coming form advertisement and movie sequences.