摘要
针对在高速动态情形下的车型识别,介绍了一种对汽车提取特征、基于红外线检测的汽车分类仪;阐述了采用汽车特征参数作为样本向量训练BP网络的方法和识别车型原理;采用共轭梯法修正BP网络,提高了训练速度和全局收敛性;对于样本向量存在的数据“噪声”,则以Bayes法则对大量样本去除“噪声”,使特征样本向量更有代表性,理论与实际证明,这样得到BP网有强容错能力,能识别没有看过的汽车样本,从而提高了车型识别精度。
A Infrared detecting Vehicle Classification machine is designed to extract vehicle characteristics in high speed condition , and a BP neural networks is adopted to train and classify the extracted characteristics. To denoise, the Bayes principle can be applied to train the mass sample, and promote the representability of the characteristic vector. The practice proved that the BP neural networks have strong capacity of error acceptance, and can classify the vehicle samples not trained. The Fast Vehicle Classification system can meet the demand of the high precision classification.
出处
《计算机测量与控制》
CSCD
2005年第7期641-644,共4页
Computer Measurement &Control
关键词
红外线检测仪
Bayes法则
BP神经网络
车型识别
infrared detecting machine
Bayes principle
BP neural networks
vehicle classification
