车牌识别系统方法研究和实现.docVIP

车牌识别系统方法的研究和实现 工程领域：软件工程 研究生 指导老师 随着社会经济的不断发展和人们生活水平的普遍提高，交通变得越来越拥挤，智能交通系统已经成为交通管理的主要方向，而车牌识别系统是智能交通管理系统的核心内容，研究车牌自动识别技术具有重要的实用价值。 论文主要完成了车牌识别系统的方法研究和设计工作。车牌识别过程主要包括：车牌预处理、车牌倾斜校正、数字字符分割和字符识别。车牌预处理方法，首先通过中值滤波对图像进行平滑处理，再用寻求最佳阈值的方法对图像进行二值化的处理。车牌倾斜校正方法，首先利用Hough变换提取车牌边框的直线，并对结果进行统计平均求出车牌的倾斜角度，再采用仿射变换得到校正后的图像。数字字符分割方法，首先利用先验知识去除边框和去除孤立单元面积，然后利用投影分割法对车牌图像进行分割，从而得到数字字符的精确位置。字符识别方法，采用基于细化图像的Hausdorff距离模板匹配识别方法，识别之前首先进行图像的归一化和细化预处理，然后建立细化图像的标准模板库，用Hausdorff距离匹配，将车牌字符识别出来。 试验结果表明，论文所提出的多种预处理与识别技术有机结合的车牌识别方法能较准确地识别车牌，整个系统的识别能力和运行性能良好。 关键词：图像去噪，二值化，字符分割，模板匹配，字符识别 License Plate Recognition Research and Implementation Field: Software Engineering Graduate Student：Han Rui Advisors：Ruan Shuhua Tang Sanping With the social economic development and peoples standard of living are general improving,traffic has become more and more congestion.Intelligent Transportation System(ITS) is the main way to solve the traffic congestion and blocking,LPR is an important field in the ITS,and it has great application value. On the basis of other peoples research，this paper has researched and designed LPR, This paper has mainly the following four parts: image preprocessing,tilt correction, character segmentation and character recognition. Image preprocessing described the pretreatment work of the image, including smooth the image by making use of the template of median filter, and binarize the image by using the method of seeking the best threshold value. Image tilt correction uses Hough transform to extract license plate line, and the results are statistical average, and then calculate the tilt angle. Finally, the image can be corrected by affine transform.character segmentation gets rid of the plate lines and sound areas through calculating the area,and then uses traditional projection to find the exact location of the characters.character recognition uses template matching to identify which is according to Hamming distance. The experiments show that the method is effective and feas