SNR scalability based on bitplane coding of matching pursuit atoms at low bit rates: fine-grained and two-layer

Abstract

Because channel capacity varies depending on network traffic and the capacity of each reception, fine granularity scalability (FGS) of video coding has emerged as an important area in multimedia streaming applications. We propose an FGS video codec and a two-layer signal-to-noise ratio (SNR) scalable video codec, based on matching pursuits and bitplane coding. The temporal and spatial redundancy of atom positions between adjacent bitplanes are explored using the quadtree representation for a bitplane and the quadtree prediction algorithm. The efficiency of encoding atom positions is evaluated. Our FGS combines successive bitplane quantization of atom modula and quadtree prediction of atom positions. The performance of our FGS is compared with that of the discrete-cosine-transform-based FGS codec. The quality of a base layer or enhancement layer video in a two-layer scalable video codec can be adjusted without changing the bit rates. We propose using a combined frame obtained from the combination of the reconstructed base layer and enhancement layer images to estimate motion vectors. The performance of our two-layer codec is illustrated.