Knowledge Engineering Perspective of Video Compression

Abstract

The focus of this paper is to explore the scope of data mining techniques in video compression. Redundant information in a video is generally classified as temporal and spatial. This superflous data takes up excessive storage space and time during transmission which can be eliminated. The proposed work incorporates various data mining techniques like dimensionality reduction, clustering and mining to prune out temporal and spatial redundancies. The results show noticeable improvement in compression ratio. The compression attained by the proposed work is in the vicinity of existing codecs.