Rate control for HEVC based on spatio-temporal context and motion complexity

Abstract

Rate control plays a crucial role in video coding. Existing rate control methods for high efficiency video coding standard (HEVC) hardly take advantage of spatio-temporal context information when updating model parameters, resulting in inaccurate computation of the Lagrange multiplier and quantization parameter (QP). Meanwhile, these methods fail to consider the motion complexity of a basic unit (BU). Therefore, a large mismatch may exist between the allocated bits and actually encoded bits. In this paper, we propose an effective rate control scheme for HEVC at the BU layer to exploit both the spatio-temporal context and motion complexity. Firstly, the actually used Lagrange multiplier and generated bits of the highly correlated coded BUs in the spatio-temporal context are carefully weighted to update the model parameters. Secondly, the allocated bits for a BU are adjusted based on its motion complexity. Thus, an improved Lagrange multiplier is more accurate in minimizing the difference between the target bits and actually generated bits. Experimental results show that our proposed method outperforms the current scheme in HM with negligible complexity increase.