基于图像处理的QR码定位及纠正算法研究

发布时间：2018-03-24 17:50

本文选题：图像识别　切入点：QR码　出处：《合肥工业大学》2016年硕士论文

【摘要】：作为应用最广泛的矩阵式二维条码,QR (Quick Response)码不仅具有容量高、纠错能力强等优点,而且可以表示汉字、响应任意角度的扫描识别,在物流管理、产品行销、电子票务等各个方面已得到大量应用。当前二维码技术的应用中,多采用人工定位,如手持扫码枪扫码、手机对焦扫码等,但是这些方法显然不适用于工作量大、内容单一的工业生产环境,如快递包裹智能分拣环节等。信息技术的不断发展,为此类需要耗费大量入力的工业生产环节提供了智能化解决方案。现有的快递包裹分拣系统多采用一维条码识别技术,随着二维码技术的成熟,其在快递包裹智能分拣系统中具有更好的前景和广泛的应用。本文针对当前二维码应用中的定位、识别困难的现象,研究常用二维码定位及识别的算法,在分析研究了这类算法优缺点的基础上,提出了一种结合符号特征对QR码进行定位及纠正的算法。该算法首先采用局部灰度最大差值采样法定位所有高对比度的区域,然后根据二维码的纹理特征和符号特征筛选二维码区域,实现QR码区域的定位,分割出整个符号以降低后续操作的运算量。根据标准QR码符号中三个位置探测图形中心点应构成等腰直角三角形的特征,通过判断拍摄所得三角形的畸变形态,确定QR码的倾斜角度并进行旋正处理,最后改进传统译码方法中采样网络校正部分以提高译码识别率。基于MATLAB开发平台对算法进行了仿真验证,结果表明,在任意角度下,该算法都能够快速准确地实现QR码的定位和识别,并且运算量低,相对于传统的边缘检测和Hough变换更加便捷准确。作者基于Microsoft Visual Studio 2010与MATLAB 2010对提出的算法进行混合编程设计,并搭建了基于二维码实时识别的快递包裹智能分拣系统,并在分拣现场针对本文提出的算法进行了实验。实验结果表明,该算法正确译码率达到95.74%以上,能够满足诸如快递包裹智能分拣环节之类的工业现场快速智能分拣的需求。
[Abstract]:As the most widely used matrix two-dimensional bar code QR QR Quick response code, it not only has the advantages of high capacity, strong error-correcting ability, but also can represent Chinese characters, respond to any angle scan and identify, in logistics management, product marketing, etc. Electronic ticketing and other aspects have been widely used. In the current application of two-dimensional code technology, manual positioning is often used, such as hand-held scanning gun, scanning code for mobile phones, and so on, but these methods are obviously not suitable for heavy workload. Single content of industrial production environment, such as the intelligent sorting of express packages, etc. The continuous development of information technology, It provides an intelligent solution for this kind of industrial production, which needs a large amount of input. Most of the existing express package sorting systems use one-dimensional barcode recognition technology, and with the maturity of the two-dimensional code technology, It has a better prospect and wide application in the intelligent sorting system of express package. Aiming at the localization and recognition of difficult phenomenon in the current QR code application, this paper studies the commonly used algorithms for the location and recognition of the QR code. Based on the analysis of the advantages and disadvantages of this algorithm, a new algorithm is proposed to locate and correct QR codes by combining the symbol features. Firstly, the local grayscale maximum difference sampling method is used to locate all the high contrast regions. Then the QR code region is selected according to the texture and symbol features of the QR code, and the location of the QR code region is realized. The whole symbol is segmented to reduce the computation of subsequent operation. According to the feature of isosceles right triangle which should be formed by detecting the center points of three positions in the standard QR code symbol, the distortion shape of the triangle obtained by shooting is judged. The tilt angle of QR code is determined and the rotation is processed. Finally, the sampling network correction part of the traditional decoding method is improved to improve the decoding recognition rate. The simulation results based on the MATLAB development platform show that the algorithm is at any angle. These algorithms can locate and recognize QR codes quickly and accurately, and the computation is low, which is more convenient and accurate than the traditional edge detection and Hough transform. The author designs the proposed algorithm based on Microsoft Visual Studio 2010 and MATLAB 2010. The intelligent sorting system of express package based on the real time recognition of QR code is built, and the experiments are carried out on the sorting site. The experimental results show that the correct decoding rate of the algorithm is over 95.74%. Able to meet the need for quick and intelligent sorting of industrial packages, such as intelligent sorting of express packages.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.4

【相似文献】