干扰线去除判断比较挫是导致识别率低的原因，希望高手指点吧~后面的特征提取和训练识别就交给大家了~ （很不幸地告诉大家，上班前12306的验证码干扰又加强了，主要还是干扰线部分）

实现原理很简单：

1.图像灰度化与 二值化

2.去除干扰线（二值化在一定程度上已经消弱部分）-Interference lines to remove judge comparative setback is the reason for the low recognition rate, hope expert guidance to behind the feature extraction and training recognition to the people to (unfortunately tell you, 12306 verification code interference strengthened before work mainly interfere with the line part) principle is very simple: Grayhound values ​ ​ of 2 to remove interference lines (binarization has been weakened to some extent, part)