Title :
Renovating flawed handwriting to improve recognition
Author :
Wang, Jianguo ; Yan, Hong
Author_Institution :
Sch. of Electr. & Inf. Eng., Sydney Univ., NSW, Australia
Abstract :
Many handwritten character recognition systems can achieve a high recognition rate in general, but yield poor accuracy for flawed handwritten characters. This paper presents a new scheme to improve the recognition of flawed handwritten characters using a sequence of connected stroke separations, broken character mending and distorted characters recognition methods associated within a hybrid recognition system. A structural recognition algorithm combined with a neural network classifier is used to test the efficiency of the proposed method for renovating flawed character. Experimental results on a large set of data show the efficiency and robustness of the proposed method for handwritten digits recognition
Keywords :
handwritten character recognition; image classification; image thinning; neural nets; broken character mending; connected stroke separations; distorted characters recognition methods; efficiency; experimental results; flawed handwriting renovation; flawed handwritten characters; handwritten character recognition systems; high recognition rate; hybrid recognition system; neural network classifier; skeleton-based structural recognition algorithm; Australia; Character recognition; Feature extraction; Handwriting recognition; Neural networks; Noise reduction; Robustness; Skeleton; Testing; Writing;
Conference_Titel :
TENCON 99. Proceedings of the IEEE Region 10 Conference
Conference_Location :
Cheju Island
Print_ISBN :
0-7803-5739-6
DOI :
10.1109/TENCON.1999.818673
