文档介绍:基于遗传算法的神经网络车牌字符识别方
法#
屈鸿,王晓斌,武丽娟*
5 (电子科技大学计算机学院)
摘要:本文将遗传算法与神经网络结合起来进行车牌字符训练和识别。根据神经网络的特
点,对遗传算法的选择、交叉和变异算子进行了改进,并通过车牌字符识别实验进行了验
证。采用实数编码的方式,提高了网络权值与阈值的计算精度,运用多种交叉方式并行交
10 叉,拓宽了物种范围,利于搜索到全局最优解,自适应的变异概率控制方式一方面保证了
物种的多样性,另一方面让算法快速收敛于全局最优。实验结果表明遗传算法提高了神经
网络的训练效果,克服了 BP 神经网络收敛慢、易陷入误差函数局部极小值的缺点,提高了
车牌识别正确率。
关键词:车牌识别;遗传算法;神经网络
15 中图分类号:
Neural network based on genetic algorithm method for
license plate recognition
quhong, Wang Xiaobin, Wu Lijuan
20 (School of Computer Science and Engineering, University)
Abstract: This paper combines genetic algorithm and neural network to recognize the license plate
characters. According to the characteristics of neural networks, we improve selection, crossover
and mutation operators of genetic algorithm, and verify by the license plate character recognition
traditional genetic algorithm is improved by using real number coding method to
25 enhance the network weights and threshold accuracy. At the same time, we use a variety of
cross-way parallel crossbar, which is broadening the range of species and