摘要： 基于笔画描述的汉字轮廓线，小波变换能更有效地提取签字的字形特征。该文提出了一种基于小波变换的笔迹鉴别的新方法，利用B-样条函数更适合于计算机图形中的曲线与曲面处理的特性，设计了适合系统特点的四阶样条小波，对签字的每一笔画进行小波变换，又设计了一组公式对各笔画的特征进行合成，从而抽取出了更加稳定有效的签字特征。

关键词: 小波变换, 样条小波, 特征值, 样条函数

Abstract: This paper presents a new approach to identify Chinese signatures based on the wavelet transform. The extracted shape eigenvalues are effective and accurate when the signature is decomposed into character strokes. The B-spline function has the significant characteristic of processing curve and surface in computer graphics. A group of 4-spline wavelet bases are proposed by B-spline function and applied in each stroke of the signature. It also designs a set of formulas to synthesize the extracted eigenvalues for each stroke so that the resulting signature eigenvalues are more effective and stable.

Key words: wavelet transform, spline wavelet, eigenvalue, spline function

中图分类号: 

• TP391.43