Optimum Bit Allocation and Rate Control for H.264 via Cauchy-Distribution-based Rate and Distortion Models

For the rate control of H.264, one of the most important things is to model the R-D characteristics accurately. To achieve this, a more accurate rate-quantization (R-Q) model and a distortion-quantization (D-Q) model were developed based on Cauchy-distribution assumption. For low bit rate application, these new R-D models were employed for optimum bit allocation by using Lagrange optimization. Experimental results show that the proposed algorithm has up to 0.79 dB gains in the average PSNR meanwhile, decreases bit estimation error significantly and control the rate more accurately, compared to JM8.6 rate control algorithm.