Matlab车牌数字字符识别算法的设计与实现
一般来说，车牌自动识别系统主要包括四个部分：车牌定位、车牌倾斜校正、车牌字符分割和车牌字符识别。车牌字符识别作为车牌自动识别系统中最后一个环节，影响着最终车牌识别的结果。在车牌识别系统中，传统的模板匹配算法处理时间长，对笔画缺失或断裂的字符识别效果不好。针对传统的单模板匹配方法适应性差、识别率低的问题，本文应用Matlab语言实现了一种基于特征融合和多模板匹配的车牌数字字符识别算法。首先对图像进行预处理操作，然后提取字符网格特征和穿越特征，并将两者融合，构成混合特征，最后基于欧式距离进行多模板匹配。其中多模板采用K-均值聚类算法获得。实验结果表明：本文算法能有效地提高字符的识别正确率，具有较高的实用价值。7402
关键词：车牌识别系统；数字字符识别；特征融合；多模板；K-均值聚类
Design and Implementation on Iicense Plate Digital Character Recognition Algorithm
Abstract:in today's rapid economic development, all kinds of automobile has become necessary in our traffic tools, with bring convenience for travel and our life, also has brought great pressure and challenges for traffic management, So intelligent transportation is becoming more and more important. As intelligent transportation is one of the most important elements of automatic license plate recognition system has become an important topic.
In general, The automatic license plate recognition system includes four parts: license plate location, license plate tilt correction, license plate license plate character segmentation and character recognition.License plate character recognition as automatic license plate recognition the last link in the system, affect the final result of license plate recognition. In the license plate recognition system, the traditional template matching algorithm, long processing time, the character recognition effect on stroke deletion or fracture is not good. For the traditional single template matching method poor adaptability, low recognition rate problems, this paper use Matlab language which is based on feature fusion and multi-template matching number plates character recognition algorithms. First pretreatment operation on the image, and then extract the character grid characteristics and through the feature, and the two fusion, consists of a mix of features, finally based on Euclidean distance for multiple template matching. Multiple templates obtained by K- clustering algorithm. The experimental results show that, this algorithm can effectively improve the recognition correct rate, has a high practical value.
Keywords: license plate recognition system; digital character recognition; multi-feature combination; multi-template; K-means
目 录
摘要 i
Abstract ii
目录 iv
1 绪论 1
1.1 车牌自动识别系统概述 1
1.2 课题的目的和意义 2
1.3 车牌数字字符识别技术国内外研究现状 3
1.3.1 我国车牌特点 3
1.3.2 常用的数字字符识别方法 4
1.3.3 数字字符识别技术发展趋势 5
1.4 本文的主要内容和结构安排 5
2 MATLAB软件介绍 7
3 数字图像处理理论基础 11
3.1 图像归一化 11