Frame based Video Retrieval using Video Signatures

The World Wide Web today has grown so wide and the video-on-demand applications and video share web are becoming very popular day-by-day on the World Wide Web. An efficient video similarity search algorithm for contentbased video retrieval is important in video-on-demand based services. However, there is no satisfying video similarity search algorithm showing cent percentage performance. It is proposed here to implement a video similarity measure algorithm based on the color-features of each video represented by a compact fixed size representation known as Video Signature. This Video signature which is based on the image signature is computed on the basis of YCbCr Histogram and the sum of its weighted means. The video signatures of videos are then used to find the similar videos in-terms of visually similar frames, by using the range. This method of similarity measure is assumed to be efficient in various aspects.