Browse Prior Art Database

Lossless, Grey-Scale Image Compression, Based on Reduced Parameter Model and Coder

IP.com Disclosure Number: IPCOM000038966D
Original Publication Date: 1987-Mar-01
Included in the Prior Art Database: 2005-Feb-01

Publishing Venue

IBM

Related People

Authors:
Arps, RB Chen, T [+details]

Abstract

A method is described for reducing the number of degrees of freedom in a Markov source model for the purpose of image compression, comprising the steps of: (1) mapping the k-component Markov states into a smaller set of single-component states by a linear collapsing function, and (2) reducing each conditional probability distribution to a few parameters by matching them to known functions with the best fit. The Markov process is a basis of a general model with memory for compressing image sources. The image is viewed as a condition pixel source (X/S) producing a time sequence of N-bit pixels: XX{xn=ne{0,1,.....(2N-1)}} A kth-order Markov source model is defined by the conditional probabilities {P(X/S)} and stationary probabilities {(S)}, where S is a Markov state consisting of k previous pels (X1,.X2,.Xj,.Xk), 1&j&k.