Browse Prior Art Database

Statistics Collection For Compression Coding With Randomizing Feature

IP.com Disclosure Number: IPCOM000048679D
Original Publication Date: 1982-Mar-01
Included in the Prior Art Database: 2005-Feb-09
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Helman, DR: AUTHOR [+4]

Abstract

This invention relates to a method for adaptively approximating a skew number from a conditioning class assumed to be a binary memoryless source as a ratio of the frequency of occurrence of the most probable symbol to that of the least probable symbol. In the prior art, Langdon and Rissanen (*) disclosed a method utilizing counters for estimating the most and least probable symbol and their probability range from a nonstationary statistical source of binary strings. The method-converted symbol counts into an index number adaptively. This was done by maintaining two counters for each conditioning state whose contents were representative of the open "past history" for that conditioning state. Upon receipt of an appropriate symbol, the object was to convert the counter values to an appropriate coding parameter.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 67% of the total text.

Page 1 of 1

Statistics Collection For Compression Coding With Randomizing Feature

This invention relates to a method for adaptively approximating a skew number from a conditioning class assumed to be a binary memoryless source as a ratio of the frequency of occurrence of the most probable symbol to that of the least probable symbol. In the prior art, Langdon and Rissanen (*) disclosed a method utilizing counters for estimating the most and least probable symbol and their probability range from a nonstationary statistical source of binary strings. The method-converted symbol counts into an index number adaptively. This was done by maintaining two counters for each conditioning state whose contents were representative of the open "past history" for that conditioning state. Upon receipt of an appropriate symbol, the object was to convert the counter values to an appropriate coding parameter. Once this was done, the counters were updated.

In the method of this invention, the step involves dispensing with the requirement that the counter or other internal variable be dependent upon the past history of symbols in any given conditioning class. The counting step is replaced by the step of updating of an initialized skew number as a function of a random number (Monte Carlo) selection process In the step, the counter is replaced by a stochastic mechanism where to each skew number there corresponds a distinct probability of a yes answer. For example, the higher the skew number, the low...