Shot boundary detection and TV commercial retrieval

Shot boundary detection is a pre-requisite process for content-based video indexing and retrieval. A robust algorithm is proposed for shot boundary detection and adopted into the TV commercial retrieval system in this paper. First of all, the HSV color histogram is used to represent visual content; Then the graph partition model is applied to construct the video continuity signal; Finally, the cut and gradual transition without monochrome frames are extracted by the neural networks optimized by PSO algorithm; the gradual transition with monochrome frames could be extracted by the character of the monochrome frames. Performance evaluation shows that our algorithm performs well with relatively high detection accuracy for different kinds of programs when we adopted the graph partition model to detect the shot boundary.