Skrupsky S. Models and methods for videoinformation compression in distributed computer systems

The thesis deals with the development of simulation models of distributed systems for videoinformation compression and video compression methods in these systems to increase their efficiency. The paper presents the analysis of methods and computational facilities for video compression. The method of distribution of videoinformation over nodes in computer system has been developed. This method reduces the bitrate of compressed videosequences in distributed system from 0,42% to 9% as compared with the method of distribution of videoinformation into equal parts (when saving quality of the result). Besides the simulation models of distributed systems for videoinformation compression have been proposed. These models allow the selection of distributed system resources. In addition the block method for motion estimation in videosequence has been improved. Proposed method, when compared with the basic method, reduces the time of videosequences compression on GPU in average 10,33 times for the SDTV format and in average 9,8 times for HDTV format. Keywords: videoinformation compression, distributed computer system, frame correlation, simulation model, motion estimation of blocks.