基于字符分割和LeNet-5网络的字符验证码识别  

Character Verification Code Recognition Based on Character Segmentation and LeNet-5 Network

作　　者：张敬勋 张俊虎[1] 赵宇波 李辉[1] ZHANG Jingxun;ZHANG Junhu;ZHAO Yubo;LI Hui(College of Information Science and Technology,Qingdao University of Science and Technology,Qingdao 266061,China;Shandong institutes of industrial technology,Qingdao 250102,China)

机构地区：[1]青岛科技大学信息科学技术学院,山东青岛266061 [2]山东产业技术研究院,山东青岛250102

出　　处：《计算机测量与控制》2023年第7期271-277,共7页Computer Measurement &Control

基　　金：国家自然科学基金项目(61702295)。

摘　　要：为了解决传统验证码识别方法效率低,精度差的问题,设计了一种先分割后识别的验证码处理方案;该方案在预处理阶段用中值滤波去噪,再利用霍夫变换对图像字符进行矫正;在字符分割阶段,利用垂直投影算法确定验证码字符块个数,以及字符坐标点,再用颜色填充算法对验证码进行初步分割,根据分割后的字符块数量对粘连字符进行二次分割;在识别阶段,我们对LeNet-5网络进行了改进,修改了输入层,并用全连接层替换了LeNet-5网络中的C5层,以此来对验证码字符进行识别;实验表明,对于非粘连验证码和粘连验证码,单张图片分割时间为0.14和0.15 ms,分割准确率为98.75%和97.25%,识别准确率为99.99%和97.7%;结果表明,该算法对验证码分割和识别都有着很好的效果。To solve the low efficiency and accuracy of traditional captcha recognition methods,a captcha processing scheme of segmentation before recognition is designed.In the preprocessing stage,median filtering is applied for the noise reduction,and Hough transform is used to correct the image characters.In the character segmentation stage,the vertical projection algorithm is used to determine the number of character blocks and their coordinates,and then the color filling algorithm is used for preliminary segmentation.Based on the number of segmented character blocks,a second segmentation for connected characters is performed.In the recognition stage,the LeNet-5 network is improved by modifying the input layer and replacing the C5 layer with a fully connected layer for character recognition.Experimental results showed that for the non-connected and connected captchas,the segmentation time for single image is 0.14ms and 0.15ms,respectively,with the segmentation accuracies of 98.75%and 97.25%and the recognition accuracies of 99.99%and 97.7%.The results show that the algorithm has a good performance for captcha segmentation and recognition.

关 键 词：字符分割 颜色填充分割算法 粘连字符 字符识别 LeNet-5网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]