Reference for citation: Rubina I. S. Algorithm of video series compression based on intra-frame prediction of pixel brightness // Izvestiya Vysshikh Uchebnykh Zavedeniy. Priborostroenie. 2015. Vol. 58, N 7. P. 520—526 (in Russian).

Abstract. Existing algorithms of intra-frame pixel brightness prediction are considered. The algorithms used in video compression standards H.265 and VP9 are analyzed, characteristics of the algorithms are studied, and recommendations on their application are formulated. It is shown that a fundamental problem in video compression is the referent frame redundancy, and reduction in spatial redundancy may be improved by increasing the efficiency of intra-frame prediction algorithms. Particular attention is paid to indicators such as quality, compression ratio and computational complexity of compression process. Improving the flexibility of the mechanism of prediction procedure selection is found to reduce significantly the computational expenses while retaining the prediction algorithm performance. An adaptive method of intra-frame prediction based on the algorithm introduced by H.265 standard is proposed. The method is invariant to variations in video scene properties and reduces significantly the computational cost at the expense of a slight decrease in the video series quality. Results of testing the developed method with the use of a number of video sequences from JCT-VC test set for the H.265 standard are presented.

Keywords: video series compression, HEVC, VP9, intra-frame prediction, adaptive mode