关键词: 自适应, 文字切分, 笔迹鉴别, 纹理

Abstract: At present，texture image for handwriting identification is commonly made by compressing row and word spaces.Long strokes randomly appearing in the sample will thus lead to uneven texture images，and then affect the result of handwriting identification.An adaptive characters segmentation algorithm based on radical merging is presented.Experiments show that the algorithm can effectively segment the conglutinate characters and create a more uniform texture image.The results of handwriting identification can be improved effectively.

Key words: self-adapting, characters segmentation, handwriting identification, texture