一种快速纸币图像污损检测算法研究与仿真  

A Fast Strain Detention Algorithm for Paper Currency Image and Morphological

作　　者：张颖[1] 韩笑[1] 

机构地区：[1]辽宁科技大学电信学院,辽宁鞍山114051

出　　处：《计算机仿真》2016年第6期442-445,450,共5页Computer Simulation

基　　金：国家自然科学基金项目(71371092)

摘　　要：为提高纸币清分系统中对纸币污损识别的效率和流通过程中细小磨损和折叠等的抗干扰能力,提高检测精度,提出了一种采用相位相关法的纸币污损检测算法。首先,利用纸币纹理、颜色、梯度信息得到纸币污损特征矩阵,进而得到污损关键区域。截取参考图像某一关键区域,根据污损特征矩阵得到待检图像对应位置并截取相同大小区域,利用相位相关法得到配准结果。然后,将配准后两图像关键区域进行像素差得到灰度偏移图像,并根据灰度偏移图像进行进一步统计处理最终得到污损检测结果。Matlab仿真结果表明,上述算法实现简单,避免了普通相位相关法中寻找匹配区域不准确且较慢的影响,大大减少了运算量,提高了效率,同时充分考虑并解决了流通中正常折叠、磨损等的干扰,为提高纸币清分系统的效率提供了依据。To improve the strain detection rates and anti - interference ability for the tiny normal wear and folding in the distribution system of paper currency, we proposed a strain detection method based on phase correlation. Tak- ing the advantage of stripes, color, information of gradients to obtain the fouling characteristic matrix of the paper currency, and then obtaining the critical areas. Capturing a key area of the reference image, according to the fouling characteristic matrix, we can get the corresponding position of the inspect image, intercept the same size, and then obtain the registration result. The gray shift image was obtained after the registration of the two images for pixel differ- ence. Further statistical processing was needed according to the gray shift image to get the result of the detention. Matlab based morphological results shows that the algorithm is easier to conduct, using the algorithm can avoid the in- fluences of inaccuracy and slowness in searching the match area, largely reduce the computational complexity, and increase the efficiency. At the same time, the interference of the normal creases in the circulation was fully consid- ered.

关 键 词：纸币图像 污损检测 特征识别 相位相关 图像配准 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]