Abstract:
Accurate statistical distribution for estimating the transformed residues is greatly important for us to analyze the rate-distortion behavior of video encoders. However, ...Show MoreMetadata
Abstract:
Accurate statistical distribution for estimating the transformed residues is greatly important for us to analyze the rate-distortion behavior of video encoders. However, the previous work pays more attention to those sequences of low-resolution. In this paper, we address the statistical characteristics of DCT coefficients of high-definition videos coded by H.264/AVC. The contribution of this paper is threefold: First, Laplacian Mixture Model (LMM) is proposed to model the residues instead of using Laplacian or Cauchy distributions; the corresponding new rate-distortion model based on LMM is presented next; based on this new rate-distortion model, one frame-layer rate control algorithm is developed. Experimental results showed that the proposed rate control method achieves an improvement of PSNR up to 0.52dB with less visual quality variation compared to JM 11.0.
Date of Conference: 19-23 July 2010
Date Added to IEEE Xplore: 23 September 2010
ISBN Information: