[This article belongs to Volume - 56, Issue - 04]
Gongcheng Kexue Yu Jishu/Advanced Engineering Science
Journal ID : AES-15-05-2024-699

Yashaswini M, Dr.K V Prasad, Dr.Hemanth Kumar A R,

Abstract : High Efficiency Video Coding (HEVC) being the contemporary video coding standard aims to impart a much finer coding performance upon comparison to that of its antecedent. Different types of techniques have been evolved over the past few years in HEVC with the purpose of enhancing coding performance. An encouraging inclination of a distributed multimedia coding is thus required that could needed, that could ease the complete multimedia research community. Much of the evolution of video communication and processing has been administered in the direction of very high-quality media services via high-efficiency video coding (HEVC). In this work a method called, Ridge Regressive Bumble Bee Mating Optimized Deep Belief Network (RRBMO-DBN) is proposed to ensure both reliable and robust multimedia data communication. The designed RRBMO-DBN method uses different types of units for identifying optimal solution towards efficient multimedia communication. The entire process is built under the deep belief network involving one input unit, two hidden units and one output unit. To start with, the population of bumble bees is considered as input in the input unit. This is performed by applying Two-directional Derivative-based Representation and Gradient-based Evaluation model. Second rate distortion cost (RD cost) and energy consumption for every CU partitioning pattern are evaluated and population initialization is made accordingly. Followed by which the fitness function is evolved for each initialized bumble bees based on the fitness function results for every CU partitioning pattern using Ridge Regression function to select optimal CU partitioning pattern for efficient HEVC video encoding/decoding on multi-core DSP processor. With optimal results, better multimedia communication is said to be achieved in terms of bit rate, PSNR, bandwidth, latency and transmission rate. The presentation of the proposed RRBMO-DBN method is evaluated in bit rate, PSNR, bandwidth, latency and transmission rate using the short videos dataset. The proposed RRBMO-DBN method has good attributes that improves the performance of in terms of PSNR by 34%, transmission rate by 18% and bandwidth by 27% respectively.