Abstract: CAPTCHA Decoder technology is used in multi-send technology.The paper presents a CAPTCHA Decoder System based on template matching.Grid feature template,crossing points feature template and improved dynamic template as well as hole number feature are used in the system.In order to improve the decoding accuracy,a weighted template is proposed and a statistical approach is adopted to obtain the weight.Tests show that the system’s decoding accuracy is 100%.

摘要： 验证码识别技术常用于网站的群发技术。使用网格特征模板、交叉点特征模板、改进的活动模板,结合孔洞数特征,设计了基于模板匹配算法的形变数字验证码识别系统。为进一步提高识别率,提出了一种加权模板及其构造方案,并运用统计学方法,通过大量随机样本的多组实验,求得各组模板及其权值。测试结果表明,系统的识别准确率达到了100%。