基于BP神经网络的汽车图片车牌识别

高速公路的不断发展和车辆管理体制的不断完善，为以图像处理为基础的智能交通管理系统进入实际应用领域提供了契机。主要是对该系统中采集的车辆图片进行车牌识别的研究，主要分为牌照区域提取、图像二值化、牌照字符分割和牌照字符识别几个步骤。其中前三步是图像预处理部分，主要为后面的牌照识别提取特征向量，核心部分的字符识别部分采用BP神经网络。通过大量数据训练调整网络权值，达到理想的识别效果。

License Plate Recognition Based on BP Neural Network

The administrative system of the intellectual traffic is a development trend of traffic administration in the 21st century. Constant development of the expressway and constant perfection of vehicle management system offered the opportunity for the traffic administrative system based on image processing. In this paper, the recognition of license plates is studied according to the vehicle pictures acquisited in the system. The process is divided into four steps： region acquisition of license plate, image binaryzation, character segmentation of license plate and character recognition of license plate. Among them, the first three steps are the section of image preprocessing, which acquisits the vector quantity of the characteristic for the license plates, the last step is the key section, which adopts BP neural network and training of mass data to adjust the weight number of the network to reach the ideal discernment result.