Analysis of color histogram based similarity search and retrieval strategy of videos in Video on Demand systems

Abstract

The advent of the internet and smart hand held devices have driven the explosion of multimedia data especially video data. It has become difficult for the end user to get his desired content in a stipulated time as services like video-on-demand systems and video share Web, the major contributors of video data, has led to the ever growing quantity of video databases. This has led to extensive research in the field of video similarity search for content-based video retrieval. Traditional methods of content based retrieval strategies are computationally expensive and do not consider the temporal features of a video. Hence a fast content based scalable similarity search strategy has been an active area of research. There are two primary challenges regarding visual similarity search problem: video similarity measure and fast search method in large database. A compact signature of video is computed according to image histogram by extracting frames of a video. The video similarity is measured by the computation of the distance of signature of video. A search method based on clustering index table by index clustering and scoring using different parameters was analyzed.