Adaptive Data Compression for Video Signals

Publisher Summary This chapter discusses the adaptive data compression for video signals. It describes the problem of encoding compressed data for transmission in a noiseless channel. The first step in the design of a sampled-data compression system is the selection of a method that will eliminate redundant samples at the transmitter and reestablish these samples at the receiver. To strictly satisfy the properties of adaptivity, the prediction operator should be updated, that is, new weights should be computed before each prediction attempt. The effects of noise on compressed data are translated into streaks or contours in the reconstructed image. The exact structure of these effects seems to be primarily dependent upon the encoding procedure chosen. At high signal to noise ratio, the run length type code provides superior coding efficiency and the noise effects become almost undetectable. The fundamental conclusion is that, using a relatively simple system, average energy savings of more than 4 dB can be achieved, and acceptable received picture quality can be maintained at channel error rates of 10 –4 and lower. The received sample error rate provides a neat quantitative measure for analysis purposes but it contains no information about the structure and distribution of the errors.