Real time video compression using DVQ and suffix trees [electronic resource] /

by Vedantam, Pavan Kumar.

Real Time Video Compression using DVQ and Suffix Trees Pavan Kumar Vedantam Video processing is a wide and varied subject area. Video compression is an important but difficult problem in video processing. Several methods and standards exist which address this problem with varying degrees of success depending on the performance measures adopted. The present research work focuses on the real- time aspect of video processing. In particular we propose a real- time video compression algorithm based on the concept of differential vector quantization and the suffix tree. Differential vector quantization is a relatively new area that focuses on efficient compression of data. The present work integrates the compression provided by Differential vector Quantization and the speed achieved by using the suffix tree data structure to develop a new real-time video compression scheme. Traditionally Suffix trees are used for string searching. In the present work, we exploit the unique structure of the suffix tree to represent image data on a tree as a DVQ dictionary. To support the special characteristics of natural images and video, the traditional suffix tree is extended to handle k-errors in the matching. The result is an orders of magnitude speedup in the matching process, making it possible to compress the video in real-time, without any special hardware. Experimental results show the performance of the proposed methodology