摘要
对手写体数字的识别问题进行了讨论,提出一种基于BP神经网络的识别方法.从而提高了识别效率.主要就在识别时,数字在图片上的位置和数字本身大小方面做了改进,发现数字在图片上的大小和其在图片上的位置直接影响识别效果.具体做的是,首先提取了图片的轮廓,然后归一化成28×28的图像.这样做,不仅使得图像数字区域大小相同,而且都在图像中心上,使得识别结果变的更加理想化,达到了高识别的目的.另外,选择了容错性较好的BP网络,以200组手写体数字图像作为输入向量,以其他的110组进行识别,效率达到了90%.
In this paper, we talk about the issue of handwritten numeral recognition, and proposes an identification method based on BP neural network. What we do is to change the location and size in the picture. We found that, they both directly affect picture reeognition effect. This paper is, firstly extracted the contour of the number, and then normalize them into the picture of. To do so, not only makes the image region the same size, but also in the image center. This makes the recognition more ideal, to achieve high identification purposes. In addition, we choose BP network with better tolerance, and 200 groups of handwritten digital image as the input vector, 110 groups of of the other as recognition, the efficiency reached more than 90%.
出处
《数学的实践与认识》
CSCD
北大核心
2014年第7期112-116,共5页
Mathematics in Practice and Theory
基金
2013年国家自然科学研究基金(61275120)
关键词
手写体数字
BP神经网络
字符识别
Handwritten number
BP neural network
character recognition
