HEVC Rate Control Optimization Algorithm Based on Video Characteristics

Abstract

Based on the relationship between the frame-level and the large coding unit (LCU)-level in the high-efficiency video coding (HEVC) rate control scheme, this paper proposes novel optimization algorithms for bit allocation at the frame-level and improves algorithms for code rate control in LCU level for the space-time domain. For the frame-level, the statistical characteristics of the whole source are considered, and the information entropy of the coded source is added to the R-lambda model. An optimization model of bit allocation considering the current coded frame is used to guide the bit allocation in the frame layer. This algorithm improves the bit accuracy in the frame level. For the LCU-level, the Hadamard transform algorithm detects the energy distribution region and motion region. It constructs a new complexity from the energy value and the predicted residual value of the image so that the target bits of the LCU layer can be reasonably adjusted and accurately assigned. The coding structure in LCU-level achieves a more accurate update of the model parameters. The experimental results show that the algorithm in this paper reduces the relative error of bit rate by 0.017% and 0.016% on average and improves the rate-distortion performance by 2.7% and 2.6% on average compared with the adaptive ratio bit allocation algorithm under low-delay B configuration and P configuration.