Handprinted Character Recognition

Abstract

This report discusses a research effort whose object is the design of a handprint alpha-numeric reader capable of human recognition rates. The effort included: Collection of a large data base of unconstrained handprinted alpha-numeric characters; Editing the data to re- label mislabeled characters, remove noise, and delete totally illegible characters. A separate programming package (the Data Base Editing Package) was designed and implemented for this purpose; The design of new features to improve the performance of the system; The design of the recognition logic using an OLPARS-like program, the Alpha-Numeric Logic Package (ANLP), which used the expanded feature set; An independent test of the logic using a set of 6127 characters not included in the design set; and, Analysis of the results of the independent test to develop reject strategies to reduce substitution errors.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1973
Accession Number
AD0755936

Entities

People

  • David B. Connell
  • John W. Sammon
  • Jon H. Sanders
  • Robert J. Mcgrath

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Abstracts
  • Algorithms
  • Character Recognition
  • Computer Programming
  • Computers
  • Data Sets
  • Databases
  • Errors
  • Feature Extraction
  • Identification
  • Information Science
  • Malfunctions
  • Measurement
  • Operating Systems
  • Personality
  • Recognition
  • Test And Evaluation

Fields of Study

  • Engineering

Readers

  • Computational Linguistics
  • Computer Programming and Software Development.
  • Computer Science/Computer Engineering/Data Science/Digital Signal Processing.