一种基于小波分解的纸币污损检测算法

A new algorithm for paper currency defect detection based on wavelet decomposition

为提高清分系统中纸币污损检测的准确率,减少撕裂和笔迹等污损对检测的影响,提出了一种基于小波分解的污损检测算法.采用仿射变换和小波变换进行纸币图像配准,运用Kirsch算子提取图像边缘信息,通过计算边缘强度差提取出纸币图像的污损特征,将纸币图像划分为若干个固定大小子区域,通过对每个区域的污损特征统计来判断该区域是否存在污损.实验结果表明污损特征对于图像灰度值相对变化具有较强的抗干扰能力,同时具有高识别率与高稳定性.

To improve the accuracy of defect detection in bank note sorting and decrease the effects of cracks and scratches of bank note on detecting, a new algorithm based on wavelet decomposition is proposed, in which affine transform and wavelet transform are applied to bank note image registration and the edge information is extracted by Kirsch operator, while the defect feature is extracted from edge intensity differential. The bank note image is divided into several fixed size subzones, in each of which the defect feature is calculated to judge the degree of contamination. The experimental results reveal that the proposed feature extraction method is robust to the gray intensity change in each subbone, and obtains high recognition rate and high stability, This method has already been used in practical bank note sorting system.