Multi-Mutative Gene Optimized High Efficiency Video Coding Using Mult-Core DSP for Reliable Multimedia Communication

Abstract

Contemporary evolution of multimedia communication escalates the insistence of high quality video coding. Nowadays High Efficiency Video Coding (HEVC) standard are instigated and embedded in most of the multimedia devices, to name a few being, mobile phones, entertainment media, gaming processors etc. HEVC standards are expected to minimize the bit rate by half for same quality upon comparison to its preceding video codec standard of H.264/AVC. However, time and space complexity for HEVC are found to be very high to get a real-time embedded solution for HEVC encoder. Multi-core Digital signal Processor (DSP) on the other hand provides a well-grounded or reliable mechanics to get rid of this drawbacks like, higher storage space, transmission bandwidth for multimedia communication and improved computational speed. In this work we plan to develop a Multi-Mutative Gene Optimized (MMGO) method of HEVC video encoder/decoder on 8 core DSP processor for efficient multimedia communication to provide progressive multi-core simulation. With the proposed MMGO method Single-instruction-multiple-data (SIMD) optimizations are utilized to optimize both the encoding time and space domain of the most consuming functions on multimedia data communication. Initially, MMGO initializes the number of genes (coding unit partitioning pattern). Next, according to the objective fitness function is determined. Finally, the optimal Coding Unit (CU) partitioning pattern of each Coding Tree Unit (CTU) is identified to enhance multimedia communication. The performance of MMGO technique is analyzed in terms of video encode/decode time, memory space, bit rate and PSNRwith recent existing reference state-of-art methods.