Block coding of graphics: A tutorial review

This paper gives a review of block coding for picture data compression. Block coding has been devised primarily for coding of graphics, but it has subsequently been extended to multilevel pictures. All the proposed codes are simple suboptimum prefix codes. Their simplicity make them suitable for real-time applications. Although blocks can be of any shape, higher efficiencies are obtained with two-dimensional blocks, thus exploiting the inherent two-dimensional correlation of pictures. According to the value of a preset parameter, block coding can be either information lossless or information lossy. In the former case, the original digitized picture can be exactly reconstructed from its coded version. In the latter case, where the compression is much higher, distortions possess easily identified features. An appropriate filtering can restore the decoded picture satisfactorily. With a slight increase in complexity, block coding can be made adaptive in a number of ways, leading to much higher compressions. For each case, comprehensive theoretical models are developed to predict the performances and to optimize the parameters. The dependence of the compression ratio on image resolution for each specific code is also examined.