利用RBF人工神经网络进行车牌识别方法的研究(论文9000字,外文翻译)
摘要：精确和高效的车牌识别方法是实现智能交通控制与管理系统的关键，而图像处理与分析、模式识别及人工智能是车牌识别方法研究与应用的重要基础。目前，如何利用图像处理、模式识别及人工智能技术提高和改进车牌识别方法的精度和效率已经成为研究的热点问题。
传统的车牌识别方法主要由四个阶段构成，即预处理、车牌定位、字符切分及字符识别，类似于传统方法，本文提出的车牌识别方法，在预处理阶段对输入的车牌图像进行灰度化，并利用OTSU法对其进行二值化；其次，在车牌定位阶段，采用数学形态学法对边缘检测后图像的车牌进行了精确的定位，然后对定位后的车牌进行校正和去边框处理以保证图像的质量；第三，在车牌识别阶段基于第一步中车牌字符的特征进行字符切分；最后，在车牌识别阶段，利用RBF人工神经网络模式识别技术作为识别的方法，对切分后的字符进行了识别。
实验结果表明，利用RBF人工神经网络进行车牌识别可以达到很好的识别效果，识别率较高，具有较高的实用价值。
关键词：车牌识别 图像处理 RBF人工神经网络
Research on license plate recognition based on RBF artificial neural network
Abstract:Accurate and efficient license plate recognition method is the key to realize intelligent traffic control and management system. Image processing and analysis, pattern recognition and artificial intelligence are the important basis for the research and application of license plate recognition method. At present, how to use image processing, pattern recognition and artificial intelligence technology to improve and improve the accuracy and efficiency of the license plate recognition method has become a hot issue.
The traditional license plate recognition method is composed of four stages, namely, pretreatment, license plate location, character segmentation and character recognition. Similar to the traditional method, the license plate recognition method proposed in this paper is used to gray the input license plate image in the pretreatment stage And then use the OTSU method to binarize it. Secondly, in the license plate positioning stage, the mathematical model method is used to locate the license plate of the edge detection image, and then the calibration and the border processing To ensure the quality of the image; thirdly, in the license plate recognition stage based on the characteristics of the first step in the license plate character character segmentation; Finally, in the license plate recognition stage, the use of RBF artificial neural network pattern recognition technology as a recognition method, After the character was identified.
The experimental results show that the RBF artificial neural network can achieve good recognition effect, and the recognition rate is high and has high practical value.
Key words: license plate recognition ；image processing ；RBF artificial neural network
目 录
一、绪论 5
1.1研究的背景与意义 5
1.2国内外研究现状 5
1.3车牌识别的难点 5
二、本文方法整体思路 6
三、本文方法描述 6
3.1图像预处理 6
3.1.1图像灰度化 6
3.1.2图像增强 7
3.1.2.1灰度变换 7
3.1.2.2图像滤波 8
3.1.3图像二值化 8
3.2车牌定位 9
3.2.1边缘腐蚀 9
3.2.2区域连通 9
3.2.3消除干扰 9
3.2.4搜索车牌 9
3.3车牌切分 10
3.3.1车牌的倾斜校正 10
3.3.1.1 Radon变换 10
3.3.1.2牌照图像的校正 11
3.3.2车牌边框的处理 12
3.3.3字符的切分 12
3.4字符的识别 13
3.4.1 RBF神经网络的基本结构 13
3.4.2 RBF神经网络的学习过程 14
3.4.3本文改进的学习算法 15
四、实验分析 16
4.1预处理结果 16
4.1.1灰度化 16
4.1.2图像增强 16
4.1.3图像二值化 16
4.2车牌定位结果 17
4.2.1边缘腐蚀 17
4.2.2区域连通 17
4.2.3消除干扰 18
4.2.4搜索车牌 18
4.3车牌切分结果 18
4.3.1车牌倾斜校正 18
4.3.2车牌边框处理 19
4.3.3字符的切分 19
4.4车牌识别结果 19
五、结论 20
参考文献： 20
致谢 22 |
