基于BP人工神经网络的车牌字符识别优化算法

Optimized character recognition algorithm based on BP artificial neural network

车牌识别是电子警察系统重要的功能模块,字符识别是车牌识别的关键步骤。目前,BP(Back Propagation)人工神经网络因其优越的性能而广泛应用到车牌识别中,但是BP神经网络在局部极值、假饱和、收敛速度缓慢等方面存在着不足。针对这些局限性,从网络的层数、节点数、动量项、学习因子方面进行分析和改进,构建了一个优化的BP人工神经网络,进行字符识别。仿真结果表明,该优化的识别算法识别准确率高,具有良好的识别性能。

License plate recognition is an important function module of E-police system. Character recognition is a key step in the process of license plate recognition. Currently, BP （Back Propagation） artificial neural network is widely used in vehicle license plate recognition because of its superior performance. However, the BP network has some disadvantages, such as the local minimum, false saturation and slow convergence. According to these draw- backs of BP networks, an optimized BP artificial neural network is built to identify characters from the aspects of the layers of network, nodes number, momentum, learning factors. The results show that the algorithm has good performance and satisfies the application required.