移动设备的信用卡号码自动提取技术  

Automatic extraction of credit card number of mobile devices

作　　者：陈佳林 CHEN Jialin(Information Institute of Ministry of Emergency Management of People’s Republic of China,Beijing 100029,China)

机构地区：[1]应急管理部信息研究院,北京100029

出　　处：《电子设计工程》2021年第11期155-159,共5页Electronic Design Engineering

摘　　要：针对智能手机在移动支付领域自动提取信用卡号码的需求,提出了基于预览视频流图像的信用卡号码自动提取算法和结合彩色Sobel梯度和LSD(Line Segment Detector)直线检测算法的透视变形校正算法;针对信用卡号码特点,提出基于Sobel梯度积分图的信用卡号码定位算法,同时提出基于SVM(Support Vector Machine)滑动窗口信用卡号码字符串自动分割算法;在字符分割基础上,通过对比梯度方向特征、Gabor特征以及对比LVQ(Learning Vector Quantization)分类器和MQDF(Modified Quadratic Discriminant Functions)分类器,采用基于梯度方向特征和LVQ分类器,设计了单字符识别引擎。实验表明,文中方法在兼顾速度的同时保持了良好的识别性能,具有较高的应用价值。A automatic extraction algorithm of credit card number based on preview video stream image and a perspective distortion correction algorithm are proposed for automatic extraction of credit card numbers of smart phones in the field of mobile payments.According to the credit card number,a credit card number location algorithm based on Sobel gradient integral graph and a automatic segmentation algorithm based on SVM(Vector Machine Support)sliding window credit card number string are proposed.On the basis of character segmentation,the single character recognition engine is designed based on the gradient direction feature and LVQ classifier by comparing the gradient direction feature,Gabor feature,LVQ(Learning Vector Quantification)classifier and MQDF(Modified Quadric Discriminant Functions)classifier.Experimental results show that the method can maintain a good recognition performance while taking into account speed,and it has high application value.

关 键 词：Sobel梯度 LSD SVM LVQ分类器 LUHN校验 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]