DFD based scene segmentation for H.263 video sequences

Content based indexing and retrieval of video is becoming increasingly important in many applications. Identifying scene changes and special effects in a video scene is an essential pre-requisite for indexing. In this paper, a sudden scene change detection algorithm for H.263 video sequences is proposed. This method uses the number of intra-coded macroblocks and displaced frame difference (DFD) signal of the video signal. Experimental results show that the performance of this algorithm is independent of the encoder threshold. Furthermore, this algorithm is capable of detecting abrupt scene changes accurately even if the video sequence contains special effects