An Enhanced Approach in Novel Content Based Video Retrieval Using Vector Quantization

Abstract: In the recent and modern world, almost everything is in the race to be digitized rapidly. Today, end-user sometimes thinks that video retrieval system based on simple text query is degrading the performance these days. So it�s a good time to move on to the content based retrieval approach based systems for searching videos. Revision of the content based video retrieval approach can lead to the effective implementation of this system. Content Based Video Retrieval (CBVR) is the best evolving system for any video retrieval application. The Block Truncation Coding (BTC) is one of the techniques in the CBVR used for color feature extraction. With improvements to the BTC, Thepade�s Sorted Ternary Block Truncation Coding (TSTBTC) is also the recent color feature extraction technique. Transform feature extraction is one more technique about extraction of the video. None of the method is right now extended for both of the features like color feature and transform feature of a video. In other hand Vector Quantization (VQ) is the lossy data compression technique. If VQ is used with TSTBTC it can deal with both of the features like color and transform feature of a video. Because the Vector Quantization (VQ) supports hybrid features (i.e. color and transform). VQ is not used before in CBVR, this paper explains the implemented stuff of VQ with CBVR.

Keywords: Content Based Video Retrieval (CBVR); Block Truncation Coding (BTC); Thepade�s Sorted Turnary Block Truncation Coding (TSTBTC); Vector Quantization (VQ); Linde-Buzo-Gray (LBG).