The compression of binary facsimile images has now become widespread and has led to the adoption of standard codes implemented in VLSI chips. In this paper, we reexamine the heuristic and formal basis for the techniques used in compressing such data. This leads us to suggest a decomposition into the successive operations of preprocessing, image segmentation or color shrinking, and modeling and coding of color blocks. Such a decomposition allows a reevaluation of the comparative performance of standard codes and leads us to propose a new simple code, the PCSE code, which, using preprocessing and color shrinking prior to encoding, outperforms the READ code by from 7 to 42% for the standard CCITT test images with a very small change in image quality.

The efficient coding of scanned documents is a problem of growing importance. The international facsimile transmission standards of 200 dpi, as well as 300 dpi high quality document scanning for computer use, are of interest. We present, and discuss, a progressive coding technique, building upon color shrinking, i.e., the gross, data dependent localization of the black portions of an image, that we have presented previously. We propose two intermediate quality standards, 100 dpi binary, and 100 dpi grey scale, which provide high compression, as well as a readable version of documents, specially suited for a computer workstation display.