QR码识别算法研究

发布时间：2018-04-03 23:51

本文选题：QR码　切入点：小波变换　出处：《南京理工大学》2017年硕士论文

【摘要】：随着信息化时代的来临,计算机科学的进步以及手机的智能化,二维条码越来越多地被应用在各行各业中。QR码作为第一种直接对非英文字符进行编码的二维条码,因其独特的编码特性和快速响应的优点,目前已经在我国的二维条码市场中占据着主导地位。然而,在日常生活中和工业生产中,采集到的QR码图像往往存在光照不均衡、扭曲形变、背景复杂等问题,在这些情况下QR码很难被正确地识读出来。因此,对QR码识别技术进行深入地研究具有十分重要的意义。QR码在识别的过程中涉及到的算法主要有图像预处理算法、定位算法和校正算法。这些算法的目的是为了提取出清晰规整的QR码符号,便于对其译码。本文深入研究了QR码识别的相关技术,并实现了针对光照不均衡,扭曲形变,背景较为复杂的QR码识别算法。首先,本文结合了 QR码符号的特点,实现了图像预处理算法。图像预处理可以分为灰度变换、光照均衡、滤波去噪和二值化这四个步骤。针对灰度变换,采用了符合人体视觉的YUV线性变换算法;针对光照不均衡,采用了简单而有效的直方图均衡化算法;针对噪声,详细研究了小波变换算法,并将其应用在QR码图像的去噪中;针对二值化,重点分析了 Otsu算法,并对其进行了改进。实验结果表明,本文实现的图像预处理算法能有效地去除图像中的干扰信息,并且极大地增强了图像的对比度。接着,本文在数学形态学以及边缘检测算法的基础上,实现了QR码定位算法,并对其进行了改进和优化。实验结果表明,改进优化后的算法能有效地除去无用的背景信息,获取完整的QR码符号,定位效果准确。然后,本文在最小二乘法和反透视变换的基础上,实现了QR码校正算法,并且针对校正图像的灰度值不连续的问题,采取了双线性插值法对其进行插值优化。实验结果表明,该QR码校正算法可有效地校正畸变的QR码,复原其真实形状。最后,研究了译码算法,并对QR码进行解读,得到了其中包含的准确信息。总体而言,本文所实现的QR码识别算法,满足了识别系统对实时性与准确性的要求,具有一定的实际应用价值。
[Abstract]:With the advent of the information age, the progress of computer science and the intelligence of mobile phone, two-dimensional bar code is more and more used in various industries. QR code is the first kind of two-dimensional bar code that directly encodes non-English characters.Because of its unique coding characteristics and the advantages of fast response, it has occupied a dominant position in the two-dimensional barcode market in China.However, in daily life and in industrial production, QR code images often have problems such as uneven illumination, distorted deformation, complex background and so on. In these cases, QR code is difficult to be correctly read out.Therefore, it is very important to study the recognition technology of QR code. The main algorithms involved in the recognition of QR code are image preprocessing algorithm, location algorithm and correction algorithm.The purpose of these algorithms is to extract clear and regular QR symbols and to decode them easily.In this paper, the related technology of QR code recognition is deeply studied, and an algorithm for QR code recognition is implemented, which is aimed at the uneven illumination, distorted deformation and complex background.Firstly, the image preprocessing algorithm is realized by combining the character of QR symbol.Image preprocessing can be divided into four steps: grayscale transformation, illumination equalization, filter denoising and binarization.Aiming at gray level transformation, YUV linear transform algorithm is adopted in accordance with human vision; aiming at illumination imbalance, simple and effective histogram equalization algorithm is adopted; wavelet transform algorithm is studied in detail for noise.It is applied to the denoising of QR code image, and the Otsu algorithm is analyzed and improved for the binarization.The experimental results show that the proposed image preprocessing algorithm can effectively remove the interference information from the image and greatly enhance the contrast of the image.Then, on the basis of mathematical morphology and edge detection algorithm, QR code location algorithm is implemented, and it is improved and optimized.The experimental results show that the improved algorithm can effectively remove the useless background information and obtain the complete QR symbol, and the localization effect is accurate.Then, based on the least square method and the inverse perspective transform, the QR code correction algorithm is implemented, and the bilinear interpolation method is adopted to optimize the QR code correction algorithm for the discontinuity of the gray value of the corrected image.Experimental results show that the QR code correction algorithm can effectively correct the distorted QR code and restore its real shape.Finally, the decoding algorithm is studied, and the QR code is interpreted to obtain the accurate information contained therein.In general, the QR code recognition algorithm realized in this paper meets the requirements of real-time and accuracy of the recognition system, and has certain practical application value.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.44

【参考文献】