Path Evaluation and Character Classifier Training on Integrated Segmentation and Recognition of Online Handwritten Japanese Character String

This paper describes a method of online handwritten Japanese character string recognition by improved path evaluation and character classifier training. The path evaluation is insensitive to the segmentation length and the optimal path can be found by dynamic programming (DP). The character classifier training improves resistance to non-character patterns, which is a problem on integrated segmentation and recognition of handwritten character strings. Experimental results show that the path evaluation and the character classifier training improve the performance of string recognition.