BP网络实现对货运列车车号识别的程序设计

THE PROGRAM DESIGN DISTINGUISHES ABOUT FREIGHT TRAIN NUMBER CHART WITH THE BP NET

基于对车号图片的分析,建立0~9十个数字的图像模型。利用MATLAB环境下BP神经网络算法在数字图像识别系统中的应用,通过对数字图像特征的提取,转置成各自的特性输入后送入BP网络训练,同时对网络的输出提出期望,使网络按照期望记忆每个数字的特性。当再输入与此数字相同或具有相似的特征的数字图像时,网络通过一定的处理程序能够自动的识别该数字。当网络具有了这种功能后,将有效的减少人工量且降低出错率[1]。

Based on the analysis of the train number image,establishes the digital picture models with the numbers 0to 9. using the application of BP nerve network algorithm under the MATLAB environment in the digital image recognition system. Causes the network according to the expectation memory each numeral characteristic. When people input the same numeral with this or the similar characteristic digital image,the network through the certain disposal procedure can recognise this numeral automatically. After the network has this kind of function,it can reduce effectively person＇s work load and also reduce the error ratio.