Enhanced Stochastic Bit Reshuffling for Fine Granular Scalable Video Coding

In this paper, we propose an enhanced stochastic bit reshuffling (SBR) scheme to deliver better subjective quality for fine granular scalable (FGS) video coding. Traditional bit-plane coding in FGS algorithm suffers from poor subjective quality due to zigzag and raster scanning order. To tackle this problem, our SBR rearranges the transmission order of each bit by its estimated rate-distortion performance. Particularly, we model the transform coefficient with a maximum likelihood based Laplacian distribution and incorporate it into the context probability model for content-aware parameter estimation. Moreover, we use a dynamic priority management scheme for the SBR. Experimental results show that our enhanced SBR together with context adaptive binary arithmetic coding offers up to 1.5dB PSNR improvement and shows better visual quality as compared to the scheme in MPEG-4 FGS.