A knowledge model based on-line recognition system

The authors present an online recognition system for 5401 handprinted Chinese characters by a knowledge model based recognition approach with stroke based features. The reference pattern is modeled with the prior knowledge of handwriting variations in stroke-order deviations and stroke-number deviations, from which a deviation-expansion model (D-E model) is constructed based on the stroke sequence of the pattern. The pattern pair or model matching with unknown pattern is converted to a matching tree on which the similarity measure function for the pattern is defined to indicate the similarity degree. The measurement of the function is made by A* algorithm based searching. The experimental results are based on 54010 sample characters in square style