Scene segmentation in video sequences by an RPCL neural network

Video database management systems require efficient methods to abstract video information. Identification of shots in a video sequence is an important task for summarizing the content of a video. We describe a neural network based technique for automatic clustering of video frames in video sequences. From each frame the features that describe the image content are extracted to form a signature. These signatures are clustered using a rival penalized competitive learning (RPCL) neural network owing its capability to being able to automatically detect the number of classes in the data set. Results presented in the paper show that for images clustering in video sequences, the RPCL network is able to automatically extract the correct number of classes, hence the correct number of scenes, and to produce a class partition which agrees with a human model of sequences.