Context-dependent audio-visual and temporal features fusion for TV commercial detection

Automatic TV commercial block detection is a key component of an intelligent commercial management system. Rather than utilizing exclusively audio-visual characteristics like most previous works, We have proposed a SVM-DP scheme to collaboratively exploit audio-visual and global temporal characteristics associated with commercials. Firstly, likelihood values of commercial and general program are calculated by context-dependent audio-visual features and SVM-based classifiers for each video shot. And then, these values are considered as observations of a two states markov chain, providing assistance for merging shots into blocks. At last, Minimum Duration Constraint (MDC) and Maximum Segment Constraint (MSC) which grasp the global temporal characteristics are presented to search optimal combination path with Dynamic Programming approaches, respectively. Experiments performed on real video data from TV channels in China show the effectiveness of the proposed scheme.