畸变二维码校正算法研究与实现

发布时间：2018-06-26 07:17

本文选题：二维条码 + 光照不均　；参考：《北京印刷学院》2017年硕士论文

【摘要】：伴随着现代信息科学技术的飞速发展,作为一项自动化信息采集技术,条码技术的应用日益广泛。在众多条码应用技术之中,虽然一维条码所涉及到的科学技术更加完善、使用到条形码的领域也更加宽广,但是一维的条形码存在着存储的信息有限、保密性低等弊端,这也使得一维条码已经不能满足社会发展的需求,因此二维条码技术应运而生。二维条码技术是以只储存有一维信息的条形码为技术基础,通过补充添加另外的一维信息所得到的二维可读条码技术,极大提升了条码的信息容量这一关键性能,这也使得条码在有限的二维平面内可以表示更多的信息。在众多二维条码技术当中,QR码具有可存储的信息量大、容错率高、可存储英文和汉字等多种文字类信息、能携带图像信息、保密性和防伪性能强等优势使得其在日常生活中得到了广泛的应用。作为二维条码技术重要的组成部分之一,二维条码识别主要采用的是图像式识别技术,首先通过图像采集设备采集含有二维条码的图像,然后经图像处理和数据处理识别二维条码,最后提取二维条码原本所包含的文本信息。根据二维条码印制在商品外包装上时遇到的光照不均和扭曲形变的现象,本文提出了畸变二维条码的校正算法。畸变二维条码的校正算法部分以数字图像处理技术为主,校正处理算法包括图像预处理、图像中二维码的定位与校正。根据图像处理的算法流程,在定位图像中的二维条码之前,首先需要对其进行预处理。在图像的色调分割处理方面,本文选取了六种不同的图像阈值分割算法,通过实验对比发现,利用全局阈值分割算法将图像作为一个整体使用同一分割阈值对其进行分割处理,采光不足的部分存在丢失有用信息的现象,虽然局部阈值分割法不存在有用信息丢失的现象,但是处理结果存在大量伪影,严重影响后续处理,于是提出了基于局部阈值分割的改进方法。为了保障图像处理后的质量,采用连通域滤波方法滤除离散于二维条码的噪声干扰点。关于QR码的位置检测方面,笔者设计了基于最小外接矩形的位置探测算法,并且通过与基于位置探测图形即回字定位图形定位的方法进行了实验对比后,采用了鲁棒性强的定位算法。在得到二维条码所在区域后,利用外包矩形对角线对二维条码进行直线逼近求出二维条码的角点和位置探测图像的角点并找出校正后图像上与其对应的点,通过计算映射关系得出校正后的图像,并提出建立采样网格,进一步提高校正质量。经实验表明,本文设计的畸变二维码校正算法能够有效提高二维条码的识别率。
[Abstract]:With the rapid development of modern information science and technology, as an automatic information collection technology, bar code technology is increasingly widely used. In many bar code application technologies, although the one-dimensional bar code involves more perfect science and technology, the field of using bar code is wider, but the one-dimensional bar code has the disadvantages of limited information storage, low confidentiality and so on. This makes the one-dimensional bar code can not meet the needs of social development, so two-dimensional bar code technology came into being. Two-dimensional barcode technology is based on the barcode which only stores one-dimensional information. By adding additional one-dimensional information, the two-dimensional readable barcode technology can greatly improve the key performance of barcode information capacity. This also enables the bar code to represent more information in a finite two-dimensional plane. Among many 2-D barcode technologies, QR codes can store large amount of information, have high fault-tolerance, can store English and Chinese characters and other kinds of information, and can carry image information. It has been widely used in daily life because of its strong security and anti-counterfeiting performance. As one of the important parts of two-dimensional barcode technology, two-dimensional bar code recognition is mainly used in image recognition technology. Firstly, the image containing two-dimensional bar code is collected by image acquisition equipment. Then the two-dimensional bar code is recognized by image processing and data processing, and the text information originally contained in the two-dimensional bar code is extracted. According to the phenomenon of uneven illumination and distorted deformation of two-dimensional barcode printed on commodity outer packaging, a correction algorithm of distorted two-dimensional bar code is presented in this paper. The correction algorithm of distorted two-dimensional bar code is mainly digital image processing, which includes image preprocessing, location and correction of two-dimensional code in image. According to the algorithm flow of image processing, it is necessary to preprocess the two-dimensional bar code before locating it. In terms of image hue segmentation, six different image threshold segmentation algorithms are selected in this paper. Through experimental comparison, it is found that the global threshold segmentation algorithm is used to segment the image as a whole using the same segmentation threshold. There is the phenomenon of missing useful information in the part of insufficient daylighting. Although there is no phenomenon of loss of useful information in the local threshold segmentation method, there are a large number of artifacts in the processing results, which seriously affect the subsequent processing. Therefore, an improved method based on local threshold segmentation is proposed. In order to ensure the image quality after processing, the connected domain filtering method is used to filter the noise interference points which are discrete to the two-dimensional bar code. Regarding the position detection of QR code, the author designs a position detection algorithm based on the minimum external rectangle, and compares it with the location method of the position detection figure, that is, the echo position figure. A robust localization algorithm is used. After obtaining the region of two-dimensional bar code, the corner point of two-dimensional bar code and the corner point of position detection image are obtained by using the outsourced rectangular diagonal line to approximate the two-dimensional bar code, and the corresponding points on the corrected image are found. The corrected image is obtained by calculating the mapping relation, and a sampling grid is proposed to further improve the calibration quality. Experimental results show that the proposed algorithm can effectively improve the recognition rate of two-dimensional barcode.
【学位授予单位】：北京印刷学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.4

【相似文献】