Saved in:
Bibliographic Details
Main Author: Humberto Sossa
Format: Artículo científico
Language:en
Published: Instituto Politécnico Nacional 2006
Subjects:
Online Access:https://www.redalyc.org/articulo.oa?id=61410306
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866812496667475968
author Humberto Sossa
author_facet Humberto Sossa
contents Associative Memories Applied to Printed Word Recognition Humberto Sossa Ricardo Barrón Benjamín Cruz Ingeniería Pattern restoration Pattern reconstruction Pattern decomposition Associative processing We describe a simple but effective methodology for therecognition of printed words from incomplete versions of them.The proposed methodology incorporates two main stages, oneof word learning and one of word recognition. During wordlearning sets of different words with different number of lettersare used to build a bank of associative memories, oneassociative memory for each group of words with the samenumber of letters. During word identification, a word isrecognized as follows: First its available letters are recognizedby means of a recently proposed recognition scheme based onpattern decomposition. Spaces between available letters arenext detected by appropriate segmentation techniques. Thewhole pattern composed of the identified letters and detectedspaces is presented to the corresponding trained associativememory that in turn responds with the appropriate restoredword. Formal conditions for correct reconstruction of a wordin the presence of alterations are provided. Numerical and realexamples are also given to show the efficiency of the proposal. 2006 artículo científico 1665-0654 https://www.redalyc.org/articulo.oa?id=61410306 en http://www.redalyc.org/revista.oa?id=614 Científica application/pdf Instituto Politécnico Nacional Científica (México) Num.3 Vol.10
format Artículo científico
id redalyc_61410306
language en
publishDate 2006
publisher Instituto Politécnico Nacional
spellingShingle Associative Memories Applied to Printed Word Recognition
Humberto Sossa
Ingeniería
Pattern restoration
Pattern reconstruction
Pattern decomposition
Associative processing
Associative Memories Applied to Printed Word Recognition Humberto Sossa Ricardo Barrón Benjamín Cruz Ingeniería Pattern restoration Pattern reconstruction Pattern decomposition Associative processing We describe a simple but effective methodology for therecognition of printed words from incomplete versions of them.The proposed methodology incorporates two main stages, oneof word learning and one of word recognition. During wordlearning sets of different words with different number of lettersare used to build a bank of associative memories, oneassociative memory for each group of words with the samenumber of letters. During word identification, a word isrecognized as follows: First its available letters are recognizedby means of a recently proposed recognition scheme based onpattern decomposition. Spaces between available letters arenext detected by appropriate segmentation techniques. Thewhole pattern composed of the identified letters and detectedspaces is presented to the corresponding trained associativememory that in turn responds with the appropriate restoredword. Formal conditions for correct reconstruction of a wordin the presence of alterations are provided. Numerical and realexamples are also given to show the efficiency of the proposal. 2006 artículo científico 1665-0654 https://www.redalyc.org/articulo.oa?id=61410306 en http://www.redalyc.org/revista.oa?id=614 Científica application/pdf Instituto Politécnico Nacional Científica (México) Num.3 Vol.10
title Associative Memories Applied to Printed Word Recognition
topic Ingeniería
Pattern restoration
Pattern reconstruction
Pattern decomposition
Associative processing
url https://www.redalyc.org/articulo.oa?id=61410306