【摘要】 提出了一种基于小波分析的笔迹鉴别改进方法。现有的利用小波分析的签字鉴别方法大都建立在把整个签字当作图像,再对整幅图像通过Mallat塔式算法进行多分辨率分析的基础上,所提取的签字特征中总体特征相对较多,细节特征较少。论文所述方法先对签字的每一笔画进行复信号小波分解,然后将反映细节特征的笔画信息进行合成,最后使用马氏距离分类器完成匹配工作。理论分析和实验结果均表明了该算法的有效性。更多还原

【Abstract】 In this paper,we describe an improved method for signature identification based on wavelet analysis.The existing methods are mostly based on that the entire signature is regarded as an image and wavelet multiresolution is carried on by Mallat pyramidal algorithm on the image,the extracted signature eigenvalues for whole signature are excessive,the particular eigenvalues are exiguous.Authors apply wavelet analysis of complex number to each stroke of signature,synthesize the extracted information of strokes as the whole signature eigenvalues and use a Mahalanobis Distance classifier to fulfil the identification task.The theory analysis Results and experiment results show this improved method is feasible.更多还原