Character recognition using neural networks

Date of Publication

1994

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Computer Science

College

College of Computer Studies

Department/Unit

Computer Science

Abstract/Summary

Artificial Neural Systems, or simply Neural Networks (NN), a technology which deals with the simulation and development of systems which exhibit some of the functions of the human brain, was introduced during the late 50's, and from then on, several applications have been developed. Speech and character recognition, loan application system, image-compression system and extraction of information from databases, are just to name a few. Among them, character recognition is, probably, the most famous, to which several developments has been made. There are several models or topologies of NN that can be applied in character recognition. Among these are Adaptive Resonance Theory, Backpropagation Network, Bidirectional Associative Memory, Boltzmann and Cauchy Machines, Counter Propagation, Neocognitron, Perceptron and Self Organizing Map. The thesis aims to investigate which among these topologies is best suited for printed character recognition. Backpropagation Network (BPN) and Adaptive Resonance Theory (ART) were the two models that were studied. Each network simulation was developed in a Turbo C++ programming environment. Together with this, pre-processing programs were also developed. Patterns used to feed the networks came from scanned images stored as PCX files. The pre-processing techniques, such as scanning, segmentation, and scaling, converted the data on the files to binary form.

Abstract Format

html

Language

English

Format

Print

Accession Number

TU07879

Shelf Location

Archives, The Learning Commons, 12F, Henry Sy Sr. Hall

Physical Description

316 numb. leaves ; Computer print-out.

Keywords

Neural networks (Computer science); Artificial intelligence; Optical pattern recognition; Computer design

This document is currently not available here.

Share

COinS