Adaptive model-based quantization for H.264 video rate control

In general, the source data to be quantized cannot be defined by a single distribution because it has a problem caused by the exhaustive approximation of the actual source distribution. In order to avoid this problem, we use the generalized Gaussian distribution (GGD) for modeling the distribution of the source. GGD is a parametric family of distributions, including the uniform, Laplacian, and Gaussian distributions as special cases. In this paper, we propose a quantization algorithm for H.264 video rate control based on the rate-quantization model, which is derived from the rate-distortion function of GGD. We implement our rate control scheme for low-delay inter-frame coding, and compare its performance with those of TM5 and TMN8 for various video sequences.