A Novel Model-Based Rate-Control Method for Portrait Video Coding

The rapid development of wireless networks and mobile devices has made mobile video communication a particularly promising service. We previously proposed an effective video form, scalable portrait video. In low-bandwidth conditions, portrait video possesses clearer shape, smoother motion, and much cheaper computational cost than discrete cosine transform (DCT)-based schemes. However, the bit rate of portrait video cannot be accurately modeled by a rate-distortion function as in DCT-based schemes. How to effectively control the bit rate is a hard challenge for portrait video. In this paper, we propose a novel model-based rate-control method. Although the coding parameters cannot be directly calculated from the target bit rate, we build a model between the bit-rate reduction and the percentage of less probable symbols (LPS) based on the principle of entropy coding, which is referred to as the LPS-rate model. We use this model to obtain the desired coding parameters. Experimental results show that the proposed method not only effectively controls the bit rate, but also significantly reduces the number of skipped frames. The principle of this method can also be applied to general bit plane coding in other image processing and video compression technologies.