Low bit rate video coding: an overview

Abstract

An overview is given of low bit rate (m*64 kb/s, m=1.6) video coding using a generalized video codec. This codec uses a hybrid coding algorithm combining differential pulse-code modulation (DPCM) in the temporal domain and transform coding or vector quantization (VQ) in the spatial domain to code sequences of spatially and temporally subsampled images. The coder uses motion detection and estimation for more efficient prediction; the resulting motion vectors are also used in the decoder for motion-compensated interpolation of the images left out in the subsampling process. All the elements of the codec are discussed and some of the options available for each element are noted.<>