Robust image and video coding with pyramid vector quantisation
James T. H. Chung-How, David Bull · 2003
Most current image and video coding standards use variable length codes to achieve compression, which renders the compressed bitstream very sensitive to channel errors. In this paper, image and video coders based on Pyramid Vector Quantisation (PVQ) and using only fixed length codes are proposed. Still image coders using PVQ in conjunction with DCT and wavelet techniques are described and their robustness to random channel errors are investigated. This work is then extended to video coding. A novel fixed rate motion compensated wavelet/PVQ video coder suitable for low bit-rate applications and generating only fixed length codewords is presented. Its compression performance and robustness to random bit errors is compared with H.263.