A combined method for stroke segmentation and feature extraction for handwritten characters
K.-M. Ku, P.P.K. Chiu · 2003
Syntactic character recognition relies on the accuracy of the processing work before the decision process. Thinning introduces distortion and noise, which makes the subsequent processes error prone. In order to increase the recognition rate, a combined algorithm is developed for processing the image in one process such that this combined algorithm need not involve a thinning process. The output of the process is in an attributed graph format. Time is saved and more accurate results are obtained. A description of the algorithm is provided.>