Browse Prior Art Database

Multilevel Image Compression Algorithm

IP.com Disclosure Number: IPCOM000053039D
Original Publication Date: 1981-Aug-01
Included in the Prior Art Database: 2005-Feb-12
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Reed, MA: AUTHOR [+3]

Abstract

An algorithm is disclosed for compressing the amount of data required to store the digital representation of an image in the memory of a computer by performing a specialized form of predictive encoding upon the results of a two-dimensional encoding of the subject image. Included is a method for generating a priori statistics on code words conditioned upon the known code words in the vicinity of the prediction. The geometry upon which the conditional probability is computed is given as well as the method for encoding correct and incorrect predictions. At each stage in the predictive encoding of code words, a probabilistic minimum length representation is achieved for each candidate code word.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 61% of the total text.

Page 1 of 1

Multilevel Image Compression Algorithm

An algorithm is disclosed for compressing the amount of data required to store the digital representation of an image in the memory of a computer by performing a specialized form of predictive encoding upon the results of a two- dimensional encoding of the subject image. Included is a method for generating a priori statistics on code words conditioned upon the known code words in the vicinity of the prediction. The geometry upon which the conditional probability is computed is given as well as the method for encoding correct and incorrect predictions. At each stage in the predictive encoding of code words, a probabilistic minimum length representation is achieved for each candidate code word.

There are several physical patterns of a priori known codes that the conditional probability can be developed to reflect, and predict the next code a posteriori. Table 1 shows two of these a priori code patterns, and Table 2 shows the resulting statistics from examination of the history line code (HLC) patterns of Table 1. Using such precomputed statistics, the set of a priori codes is used to predict the successor code. In theory, as long as no errors are made in prediction, then no codes need to be stored. As a subsequent code is predicted, the window upon which the conditional probabilities are predicated is moved forward and another prediction is made. See original.

In the dual-level image compression algorithm, if the decision is c...