A Pattern Deformational Model and Bayes Error-Correcting Recognition System.
Abstract
Various types of pattern deformations are investigated from the syntactic point of view and categorized into two major types: local deformations and structural deformations. Random noise, distortion variations, and substitutions, of pattern primitives belong to the former; syntactic errors due to pattern structural changes, such as primitive deletions and insertions, belong to the latter. Every observed pattern can be regarded as transformed from a pure pattern through these two types of deformations. An error-correcting parsing scheme for local deformations optimum in the Bayes sense is proposed. Two structure-preserved error-correcting parsers, one for string languages, the other tree languages, are also presented. Finally, further researches concerning error-correcting parsings for structural deformations and a complete error-correcting systems for both kinds of deformations are suggested.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1978
- Accession Number
- ADA058791
Entities
People
- King Sun Fu
- W. H. Tsai
Organizations
- Purdue University