Implementation of Dynamic Time Warping for Video Indexing

This paper presents implementation of dynamic time warping method (DTWM) to identify a video frame from a set of videos. The input for the search of a frame is speech uttered by persons. DTWM is applied to audio contents of various frames. The length of audio information will not be same at different instances. In such situation DTWM approach will be much better compared to template method. In order to improve the performance of DTWM, Radial basis function has been used. The performance of the proposed system is based on the quality of the audio stored inside the frames.