建筑工程中手写体常用数学符号的神经网络识别

RECOGNITION OF HANDWRITTEN MATH CHARACTERS IN CONSTRUCTION INDUSTRY BY NEURAL NETWORK TECHNIQUE

本文对建筑工程中实用的字符集识别提出了一种传统方法与神经网络相结合的新技术.在本文的实验中采用两个不同的网络:标准的三层BP网络和带一个高斯隐层的三层网络,互补地识别了建筑业中常用的二十个字符.

This paper studies a. new technique for recognition of handwritten characters in building industry, including ten numlerals and ' + ', ' - ' etc, the other ten characters. The neural network is combination of the standard BP network and a three-layer network with one Gaussian hidden layer. The article also shows a good result and the importance to constrution industry.