The performance of an adaptive image data compression system in the presence of noise

In the present paper the problem of image data compression is considered, and the evaluation of system performance with quantitative figures of merit as well as qualitative results is attempted. Three adaptive prediction techniques proposed by Balakrishnan are described and applied to video data for the removal of so-called redundant samples. Typical prediction efficiencies of several predictors are cited and compared. Encoding of "compressed" data for transmission in a noiseless channel is next considered. The efficiencies of several practical encoding procedures are compared with the efficiency of the ideal noiseless code. The most important results describe the performance of an image data compression system in a noisy channel. Davisson has related the sample error rate of the data reconstructed at the receiver to the channel error rate for a special case. This result is used as a quantitative basis for evaluating the effects of channel errors on the performance of a specific compression system. In addition, sample television pictures which depict the qualitative effects of channel errors are exhibited. Finally, an experimental laboratory facility is briefly described, and its use in this and future work is outlined.