摘要
车型识别具有广阔的应用前景,BP神经网络在车型识别中能够提高车型的识别率。在任何车型大致都可以抽象成一个"工"字型情况下,提取其中的顶长比、前后比和顶高比这三项相对参数作为BP神经网络的输入参数。采用三层3-8-3的BP神经网络,并用14对输入参数离线训练,再用4对新数据进行检验,均得到了预想的期望值。
Vehicle type recognition has a wide future application prospect.BP artificial neural network can improve the recognition rate of vehicle type in the vehicle type automatic recognition system.Any vehicle type can be Abstracted into a ＂工＂-shaped mode.Three relative parameters are used as the input parameters of BP ANN.A three layered 3-8-3 BP ANN is adopted and is trained by 14 pair input parameters.The BP ANN is tested by 4 pair input parameters and then some expected values are obtained in this processing.
出处
《计算机与现代化》
2012年第5期20-22,共3页
Computer and Modernization
