基于人工神经网络的手写体数字识别方法

数字识别是人工神经网络中的一项基本又重要的应用研究领域。为此提出了一种用简单BP神经网络识别手写体数字的方法。利用BP神经网络的良好监督学习功能,并结合提取的降维数字字符图像的灰度特征进行网络训练,提出了一种基于统计特征分类的手写体数字识别方法。实验结果表明,该方法简单且有较好的识别效果。

Handwritten numerical recognition method based on BP neural network

Numerical recognition is an important and basic part of applying technology study in artificial neural network.A simple neural network approach is proposed to give an available recognition method for handwritten numeral characters in this paper.BP neural networks which has good supervised learning function,when optimized with dimension reduction for the image feature,was applied to classify the characters of the statistical features.Experiments indicated that the simple method works well in recognizing the Arab numbers,exactly and quickly.