COMPRESSION OF GREY DIGITAL IMAGES USING GREY SEPARATIONS

Publishing Venue

Xerox Disclosure Journal

Abstract

Future Electronic and System Reprographic imaging products will be capable of rendering grey images, that is, images which specify a particular grey level at each spatial pixel site. Multi-bit per pixel images will typically fail to compress when processed with standard compression techniques like those specified by CCITT. Furthermore, because the size of these images (as measured by the total bits needed to describe the image) is substantially larger than the equivalent binary image, the processing bandwidth needed to compress an image gets much larger.

XEROX DISCLOSURE JOURNAL

COMPRESSION OF GREY Proposed Classification DIGITAL IMAGES USING GREY
SEPARATIONS Int. C1. H04n 1/41 William T. Crocca
Robert E. Coward

U.S. C1.358/429

Future Electronic and System Reprographic imaging products will be capable of rendering grey images, that is, images which specify a particular grey level at each spatial pixel site. Multi-bit per pixel images will typically fail to compress when processed with standard compression techniques like those specified by CCITT. Furthermore, because the size of these images (as measured by the total bits needed to describe the image) is substantially larger than the equivalent binary image, the processing bandwidth needed to compress an image gets much larger.

Proposed is a technique for separating a grey image into a small number of binary images which are then processed individually. The advantages to this approach are two. First, the resulting binary grey separations will successfully compress where the original grey image would not. Second, both the compression and/or decompression on each separation (or sub image) can occur in parallel, thus enabling significant improvement in processing bandwidth, while still using existing hardware implementations.

The first step employed in this compression strategy is to separate the target grey image into a fixed number of binary images. There are two distinct methods which could be used. The first is Binary separation wherein the image could be separated into binary images each of which represents a specific level of grey. There are two disadvantages to this kind of separation: first an image represented with 'n' bitdpixel will produce 2n-1 separations. The result is significant data explosion. Second, this method required interpreting each pixel description as a number, thus creating a fairly high computational cost to extract each separation. The second method for separating the grey image into a binary image is by Grey separations. The image is separated into grey magnitude images which represent the equivalent radix component of the grey scale. This kind of separation avoids the disadvantages sited in the Binary separation method. First, an image represented with 3-1' bi...