A New Quantization for Rate Control with Frame Variation Consideration

The main role of rate control in video coding is to handle a trade-off between a bandwidth limit and a video quality. In the low-delay video communication systems, its responsibility is more important. In this work, we propose a new rate control algorithm. In this paper, we have two approaches: one is a consideration of variations among the frames at the frame-level bit allocation, and the other is a model-based adaptive quantization at the macroblock-level. The proposed scheme is done with a low computational complexity. For simulations, we implement the proposed algorithm to H.263 video codec since it is widely used as a video coding tool in real-time mobile AV systems. The proposed scheme shows more improved picture quality than H.263 TMN8 and generates coding bits close to the target bits. The results also show more regular PSNR fluctuation than H.263 TMN8