A QUICK THINNING AND SEGMENTATION ALGORITHM FOR HAND-PRINTED CHINESE CHARACTERS

The paper proposes a quick thinning and segmentation algorithm for hand-printed Chinese characters going on simultaneously. The memory space occupied by the character image data is only one four hundredth of that by the normal character matrix. Any hand-printed Chinese characer can be thinned and segmented into several simple lines,are and corner strokes. It is very efficient for structual feature extraction for Chinese characters. A new chain code for edge description is also proposed, whose average chain code length is about 20% less than that of Freeman chain code.

Project Supported:

Clc Number:

TP391.4