Research on Wooden Blocked Tibetan Character Segmentation Based on Drop Penetration Algorithm
Ngodrup, Dongcai Zhao · 2010
In the field of character recognition, word segmentation has been considered as a part of pre-processing of the character recognition. The segmentation of those adhered or hidden characters has particularly been focused as one of the most difficult techniques in character recognition. The most characters appeared in wooden blocked Tibetan script are curved based on handwriting, and then printed on paper. In order to segment the each character rapidly and accurately, we presented a new method here to segment the characters based on Drop Penetration Algorithm after summarizing previous methods and experience. The new method is to insert water molecules into the image, and then identify the existent or disappearance of the text points according to force situation that water molecules pressed to text points in the image, and finally to complete the segmentation of the entire text.