Segmentation Of Off-line Handwritten Chinese Characters Based On Structure Analysis

Off-line handwritten Chinese character recognition is complex in structure and more freely for hand-writing. It is a great item which is valuable in research and prospective in application. Its recognition rate is closely related to its segmentation accuracy and the fault of segmentation definitely leads to the wrong recognition. So the research of off-line handwritten Chinese character segmentation technique is significant to its recognition technology's development.This article proposes an algorithm which is based on the structure analysis segmentation in view of the off-line handwritten Chinese character recognition. Firstly, the algorithm realizes non-adhesion off-line handwritten Chinese character segmentation using the structure information and the merge confidence level of off-line handwritten Chinese character; and then designs a connected segmentation method based on structure cluster and strokes analysis for the adhesion of handwritten Chinese characters. Through a large number of document images from different people, experiments show that the design of off-line handwritten Chinese character segmentation method is able to achieve a relative good segmentation result.