基于小波神经网络的胎号识别算法研究

Study of the recognition of tyre number based on wavelet neural network algorithm

针对人工神经网络的特点,对传统BP算法进行了改进。采用小波神经网络方法,有效克服了传统BP算法在实际应用中学习收敛速度慢和容易出现局部极小点的缺点。以轮胎胎号字符识别为例,分别用投影法和Hu不变距方法进行特征提取,并将所提取的特征用作神经网络输入层的神经元。将所设计的小波神经网络经训练后用于胎号的识别。仿真结果表明,小波神经网络在字符识别方面是一个十分有效的方法。

Aimed at neural network characteristics, the traditional BP algorithm has been improved as for its characteristics. The wavelet neural network （WNN） is applied. The shortcomings of the traditional BP algorithm, such as slow learning convergence-speed and being easy to appear minimum partly, have been overcome by wavelet neural network. In the background of character recognition of tyre number,the projection method and Hu invariant moments are adopted to extract the features which are used as the neurons of the input layer. After being trained, the WNN is used in the recognition of tyre number. The simulation results show that the WNN is a quite effective method in the character recognition.