Simultaneous MAP-Based Video Denoising and Rate-Distortion Optimized Video Encoding

Abstract

In this paper, a simultaneous MAP-based video denoising and rate-distortion optimized video encoding algorithm is proposed. We begin with formulating the denoising problem as a maximum a posteriori (MAP) estimate problem. Then, according to the Bayes rule, we show that the MAP estimate is determined by two terms: noise conditional density model and priori conditional density model. Based on the assumptions that the noise satisfies Gaussian distribution and the priori model is measured by the bit-rate, the MAP estimate can be expressed as a rate distortion optimization problem. With this, we are able to simultaneously perform MAP-based video denoising and rate-distortion optimized video encoding under some assumptions. Moreover, we describe in details how to select suitable coding parameters, i.e., quantization parameter, mode, motion vector, reference index, and regularization parameter. Finally, we conduct several experiments to verify our proposed algorithm.