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 e

程序是对手写体数字进行识别，用knn算法，代码编写是用java。训练样本和验证样本都在代码中提供了。-The program is to handwriting figures to identify, with knn algorithm, the code is written with java. Training samples and validation samples are provided in the code.

用java编写的手写体数字识别，采用knn方法，识别的训练和测试对象来自mnist数据库的数据，已经将解压后文件放进去了，算法包括文件的读取，测试部分还有识别算法。 -Written by java handwriting digital recognition, using knn method to identify the training and test objects the mnist data, has been extracted after the file into t