摘要
首先对三种车型进行特征提取,以提取的数据作为神经网络的输入,并且采取了感知机识别,BP网络识别,径向基网络识别,最后采用多神经网络进行识别。仿真实验结果表明:采用多神经网络分类器融合的方法比单一神经网络识别率高,这对提高目标识别的准确性是十分重要的。
In order to recognize an auto s type,features of three sorts of automotive types were extracted firstly and then the extracted data were used as the input of neural network.Recognitions of perception neural network,BP neural network and radial group neural network were made to identify automotive types.The simulation results show that the multi-neural network is more efficient than single neural network,which is very significant for improving target recognition.
出处
《辽东学院学报（自然科学版）》
CAS
2007年第3期135-138,共4页
Journal of Eastern Liaoning University:Natural Science Edition
关键词
神经网络
特征提取
目标识别
neural network
feature extraction
target recognition
