数字识别是字符识别中的一个重要的研究课题。阿拉伯数字只有0～9十个,而且笔画简单,但是识别这十个数字并非易事,识别的正确率并不高。虽然液晶数字在形式上相对于识别难度较大的手写体数字来说比较规范,但是液晶数字的变体也比较多,形状各异,因此文章研究的目的在于找到一种对于液晶数字识别的一般的有效的方法。文章建立了＂仪表监控与数字识别系统＂,主要对液晶数字识别的过程进行了讨论,重点研究了数字图像处理和数字识别,做了以下几个方面的工作：1）在研究了多种液晶数字特征的基础上,提取了数字的粗网格特征。2）建立了一个基于BP神经网络的液晶数字识别系统,进行了分类器设计,并对数字识别结果进行了检验,取得了较好的系统性能,从而证明了采用的方法是可行的。%Digital recognition is a very important project in the research of character recognition. The Arabic numbers only have ten numbers, ie from 0 to 9, and the handwriting is simple. But it＇s not easy to identify these ten numbers. The ratio of identification is not high. Though the digital number in LCD is more formal than the handwritten number, the digital numbers still have a lot of variants, and their shapes are different. The purpose of this thesis is to find an effective method of LCD number recognition. [n the master thesis, the au- thor establishes ＂the instrumental monitor and digital recognition system＂. The thesis mainly discusses in the process of LCD identification, especially studies the digital image process and digital identification. The author mainly does the following work： 1. Based on the study of many kinds of LCD, the author extracts the rough grid feature. 2. Establishes an LCD identification system based on the BP neural network, designs the classifier, tests the results of the classifier, and achieves good performance, which proves the method of the thesis is feasible.
展开▼
