基于Android平台的车牌识别系统的研究

发布时间：2018-01-17 18:01

本文关键词：基于Android平台的车牌识别系统的研究　出处：《江苏科技大学》2016年硕士论文　论文类型：学位论文

【摘要】：伴随着我国社会经济水平的不断提升,物联网这一产业日益发挥着其巨大的影响力。作为其融入日常生活的前沿应用之一,智能交通给我们带来了前所未有的便利。车牌识别的技术研究一直是智能交通领域的热门方向,已经在固定终端平台上得到了全面的发展。而随着智能手机的逐渐普及,车牌识别系统的便携化、多功能化也将成为其下一步的发展趋势。本文对车牌识别系统中的图像预处理、车牌定位、字符分割以及字符识别等相关操作过程进行了深入的探讨和研究,继而提出了一套完整的基于Android平台移动终端的车牌识别系统。相关主要内容如下:(1)根据实际拍摄环境的不同易造成车牌图像光照不均匀和图像过暗等现象,本文提出了一种合理的预处理改进方法。同时结合移动平台固有的硬件特点,本文提出了一种融合了边缘检测、形态学处理和投影分析的有效快速的定位方法。该方法通过sobel垂直算子对图像的边缘信息进行检测,通过形态学一系列处理运算和筛选从而将车牌连通区域分离出,达到初步定位的效果,最后结合水平和垂直投影的分析,准确地于车牌上下左右边缘处进行切割,从而精确定位出完整车牌。最后,本文针对手机拍摄车牌照片时存在远近不一,导致车牌目标区域过小影响定位效果的情况,实现了一种基于高斯尺度变换的改进方法。(2)根据实际情况考虑,本文提出了一种通过提取车牌中字符边界来进行倾斜校正的方法,根据提取的字符边界划定连线,通过斜率以计算目标车牌的倾斜角度,对于车牌边框残缺不全的情况具有很好的性能。同时研究了一种根据先验知识与垂直投影进行字符分割处理的方法,在具体实验中对字符粘连的现象也有一定的改进。(3)本文结合了模版匹配简单快速的特点,使用改进的模版匹配方法。通过给相应模版加高斯噪声的方法模拟现实拍摄环境,提高了识别算法的鲁棒性。对于相似字符的识别,本文又提出了对相似字符进行差别分析的方法进行相关处理。(4)将各部分算法移植应用于Android平台,实现完整的便携式车牌识别系统。
[Abstract]:With the continuous improvement of the social and economic level of our country, the Internet of things industry is increasingly exerting its enormous influence, as one of the frontier applications of its integration into daily life. Intelligent transportation has brought us unprecedented convenience. The research of license plate recognition technology has always been a hot direction in the field of intelligent transportation. Has been a comprehensive development in the fixed terminal platform. But with the gradual popularity of smart phones, the license plate recognition system is portable. Multi-function will also become its next development trend. In this paper, the image preprocessing, license plate location, character segmentation and character recognition in the license plate recognition system are deeply discussed and studied. Then a complete license plate recognition system based on Android platform mobile terminal is proposed. The main contents are as follows: 1). According to the actual shooting environment, the license plate image illumination is uneven and the image is too dark. In this paper, a reasonable preprocessing improvement method is proposed. At the same time, considering the inherent hardware characteristics of the mobile platform, this paper proposes a fusion of edge detection. An efficient and fast location method for morphologic processing and projection analysis, which detects the edge information of the image by sobel vertical operator. Through a series of morphological processing operations and screening to separate the connected area of license plate to achieve the effect of initial location, and finally combined with horizontal and vertical projection analysis. In order to locate the complete license plate accurately, the license plate is cut from the left and right edges of the license plate accurately. Finally, the paper takes the license plate photos on the mobile phone, which is far and near. Because of the situation that the license plate target area is too small to affect the localization effect, an improved method based on Gao Si's scale transformation is realized, which is considered according to the actual situation. In this paper, we propose a method of skew correction by extracting the character boundary of the license plate. The line is delimited according to the extracted character boundary, and the slope is used to calculate the tilt angle of the target license plate. It has good performance for incomplete license plate border. At the same time, a method of character segmentation based on prior knowledge and vertical projection is studied. In the specific experiment, the phenomenon of character adhesion is also improved. 3) this paper combines the characteristics of template matching simple and fast. The improved template matching method is used to simulate the real shooting environment by adding Gao Si noise to the template. The robustness of the recognition algorithm is improved. In this paper, a method of differential analysis of similar characters is proposed to deal with each part of the algorithm. The algorithm is transplanted to Android platform to realize a complete portable license plate recognition system.
【学位授予单位】：江苏科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.41;TP316

【相似文献】