An online recognition algorithm of handwritten Uyghur characters

This paper presents a recognition algorithm of handwritten Uyghur characters based on the statistical and structural features. According to the number of strokes, the characters are divided into five subsets, and then each subset is further split into smaller subsets based on special features like “loop”, “the position of additional strokes” and so on. A SVM classifier is designed for each subset in order to train and recognize.