Rate Control Algorithm For High Efficiency Video Coding (HEVC)

With the rapid development of 3G technology and Internet technology, the HEVC(High Efficiency Video Coding) studied by the JCT-VC is replacing the H.264/AVC due to its high coding efficiency and suit for high definition videos, being the new generation of video coding standard. In practical application, how to guarantee the quality of coding under the limited bandwidth is a very important problem because the bandwidth resource is limited and scarce, this is what rate control technology studies. For different applications, the requirement for rate control is different. In entertainment, the high-definition and quality is more important, and in real-time system like video conference, the low delay is required. The rapid development of real-time video application makes the low delay rate control a hot point in research. In this paper, we improve the existed algorithm in HEVC, and make it suitable for the application with low delay.This paper analyzes the technology in HEVC and the theory of rate control and makes a summary. Because of the weakness in low delay application of the rate control algorithm proposed in JCTVC-K0103, which is applied by HEVC standard test model(HM), this paper makes some improvements for the bit allocation scheme. In frame level, a bit allocation scheme based on the buffer status is proposed. In the coded tree unit(CTU) level, based on the R ??(Rate-lambda) model, the bit for each CTU not encoded is allocated according to a new weight and the buffer status together. The experimental results show that the bit allocation after considering the buffer status makes the rate control more accurate, and solve the problem of buffer overflow and underflow, which achieves the goal of low delay. At the same time, the R ?? model is not accurate for Intra frame rate control, which makes the performance worse when scene-cut happens, so in this paper, a frame(CTU) complexity based gradient is proposed, and use the gpp(gradient per pixel) optimize the R ?? model for Intra frame rate control. And lots of experiments show that the bit allocation for Intra frame is more accurate after the improved R ?? model is applied in Intra frame rate control, while keeping a smooth frame bit and a stable video’s quality.