Segmentation/Recognition of Hand-Written Numeral Characters

Authors

Khalid Sherdil, Washington University in St Louis

Document Type

Technical Report

Department

Computer Science and Engineering

Publication Date

1993-01-01

Filename

WUCS-93-28.PDF

Technical Report Number

WUCS-93-28

Abstract

This thesis describes a number of techniques for segmenting non-cursive handwritten digits into individual characters. It strongly emphasizes on a recognition-segmentation algorithm, which uses the linear regression method to recognize those strokes which consist of one or more straight-lined parts. A new method of sampling the pen data according to the pen speed, hence giving a more uniform points concentratino distribution, is also introduced. It is shown how several of our segmenting techniques, such as relative stroke lengths, relative stroke positions, order of stroke entry, stroke direction, stroke intersection, etc. can be combined to yield success results of about 95%.

Comments

Permanent URL: http://dx.doi.org/10.7936/K75B00QN

Recommended Citation

Sherdil, Khalid, "Segmentation/Recognition of Hand-Written Numeral Characters" Report Number: WUCS-93-28 (1993). All Computer Science and Engineering Research.
https://openscholarship.wustl.edu/cse_research/316