中文签名图像的笔划归一化

Stroke Normalization of Chinese Signature Image

根据签名图像预处理应遵循的基本原则,对目前中文签名图像预处理中存在的问题进行了详细分析,并提出了相应的解决方法.着重研究了签名图像的归一化,提出用基于数学形态学的笔划粗细归一化代替目前常用的细线化.实验结果表明,该方法既消除了因书写工具的不同而造成的差异,又保留了作者签名时笔划的细节.

Defects existing in normal pre-processing are analyzed on the basis of signature image processing principles, and the improved methods are proposed. Image normalization is researched especially, and stroke normalization is proposed to replace skeleton. It has proved that the differences made by tools can be removed while the details of strokes can be reserved.