Title :
Handprinted digit recognition by stroke tracing
Author_Institution :
Dept. of Comput. Sci., Calgary Univ., Alta., Canada
Abstract :
A structural method of recognizing handprinted digits is described, in which the strokes comprising the digits are decomposed into line segments. The lines are matched against stroke templates, beginning at a “pen down” point and traced as far as possible. Both direction and length are important to the template matching. Error rates of 3.3% are reported
Keywords :
character recognition; handwriting recognition; pattern matching; error rates; handprinted digit recognition; line segment matching; pen-down point; stroke decomposition; stroke direction; stroke length; stroke templates; stroke tracing; template matching; Computer science; Computer vision; Fingers; Laboratories; Particle measurements; Pattern matching; Pattern recognition; Shape; Software testing;
Conference_Titel :
Intelligent Information Systems, 1995. ANZIIS-95. Proceedings of the Third Australian and New Zealand Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-86422-430-3
DOI :
10.1109/ANZIIS.1995.705716
