Integrated technique with neurocomputing for temporal video segmentation
ABSTRACT Partitioning a video source into meaningful segments is an important step of video indexing. Many algorithms have been proposed for detecting video shot boundaries and classifying both shot and shot transition types. Different methods are suitable for different situations and most of the existing methods consider a threshold value determining the boundary between the two shots. However, selection of a generalized optimal threshold value is an extremely difficult task. In this paper, we propose an integrated method based on one of the popular soft computing techniques, namely neurocomputing, for temporal video segmentation that avoids problem with threshold calculations. We used a feedforward neural network trained using backpropagation algorithms. The soft computing model was trained using 80% of the frames data and the remaining 20% was used for testing and validation purposes. A performance comparison was made among the proposed soft computing method and traditional methods namely histogram difference, DCT difference, and motion difference, for temporal shot detection.