【作者】 李峰；

【导师】 徐长青；

【作者基本信息】 吉林大学 ， 软件工程， 2014， 硕士

【摘要】 车牌中包含了上路行驶汽车的身份信息，对于行驶的汽车而言是必不可少的部分。近些年，随着中国经济的发展，以及中国人口众多，中国汽车的数量上升越来越快。汽车数量的增加给汽车管理部门带来不小的压力，以往依靠人力记录登记汽车车牌的方法已经跟不上时代的发展，而且，人工效率低、速度慢。这使得车牌自动识别系统重要性越来越大,车牌识别也是近年来国内外研究的热点方向。本文介绍和分析了国内外目前的研究成果，在此基础上，文中探讨了车牌识别涉及到的各个模块的技术，设计实现了车牌识别系统。本文首先对车牌图像预处理系统进行介绍。对于车牌识别而言，预处理对系统十分关键，良好的系统性能依靠预处理得到的准确信息。所以对预处理在图像的应用进行了全面的分析。首先将摄取到的彩色图像转换为合适的灰度图像，接着对灰度图像处理，该过程包括滤波及去除噪声，然后把灰度图像的对比度增强，最后再将图像进行二值化处理。车牌定位技术也是本系统不容忽视的重点，文中阐述了车牌的特征以及常用的车牌定位技术，并对这些技术进行比较，提出了扫描定位车牌的方法；基于车牌字符的特征，完成了车牌字符分割；最后介绍了流行的字符识别技术，阐述了文中采用的字符匹配识别方法，提高了系统的准确度和速度。本文重点分析车牌区域的定位和字符分割与识别。车牌区域定位作为汽车牌照识别系统中非常关键的环节，本文先通过处理得到二值图像的边缘点图，减少噪声干扰后，根据车牌区域的竖向纹理特征初步选定候选车牌区域，然后再根据车牌的一般大小比例特征对候选车牌区域进行最后的精确定位，定位的准确率比较高。字符分割与识别：首先对得到的车牌进行倾斜度调整，接着通过分析字符的特征对字符进行分割得到独立字符，为了能更准确地进行字符的识别，对字符大小尺度进行归一化统一到一致大小。最后采用一种适应性较强的特征提取法提取字符特征，并结合了模板匹配，对七个车牌字符分别进行识别，识别的准确率较高。基于上面的工作，设计了完整的车牌识别系统，对整个系统进行测试和分析。通过实践可以看出，车牌识别系统能够快速准确地把车牌区域进行定位，同时把车牌中的字符有效地识别出来。在未来智能化交通的发展中，本系统有很高的应用潜能。更多还原

【Abstract】 The license plate includes the identity of cars on the road as an essential part forthe driving cars. In recent years, the number of Chinese people is large, and thenumber of vehicles has been rising faster and faster with the economic fastdevelopment of China. The increasing vehicles bring the management department alot of stress. The way recording the vehicle license plate by human can not keep upwith the development of the times. And human’s efficiency is low, and velocity isslow. License plate recognition is more and more important. The study direction hasbecome the topic for License plate recognition. This paper introduces and analysesthe present research results at home and abroad. Based on this, the various modulesof this system have been explored and the design has been implemented.At first, this paper introduced the license plate pointer preprocessing systemwhich is the very essential to the system and affects the performance of the system.So the comprehensive analysis has been carried on for this. Firstly, the gained colorpointers need to be converted into gray to be further processed. Secondly, graypointers are to be filtered and removed noises. Then the contrast of the pointersshould be enhanced. Last step is point binarization. License plate location is the keytechnology in the system, which can not be ignored. This paper introduces thecharacteristics of plates and the plate location technology of commonly usedand made comparison of them. The methods of scanning and locating are put forward,and character segmentation is finished by the characteristics of license plate. Finallythe popular character recognition technology has been studied with more accuracyand faster speed of the system.License plate location and character segmentation and recognition have beenstudied as an emphasis. License plate location is a crucial step in the License PlateRecognition System. This paper takes the image based on the edge point, and itlocates the license plate image according to the vertical texture feature of license plate region. This method can give a satisfactory recognition effect. Charactersegmentation and recognition: First, the method of inclination rectification is appliedto license plate image. Second, the corresponding connection region is used todivide up and get the isolate character. Third, the characters are normalized, preparingthe characters of the same size for the next character recognition. And then incharacter recognition section, feature extraction method can be chosen with strongadaptability, select many feature points, and combine with the template matchingmethod, to identify the character of Chinese characters, letters and numbers quickly, ahigh recognition rate and reliability are obtained.Based on the above，the system software has been designed. Test and analysis ofthe system have been done. This system can locate the license plate region andidentify the character on the license plate fast and accurately. In the development ofintelligent transportation in the future this vehicle license plate auto recognitionsystem has very high potential application.更多还原