Research On Recognition Of QR Codes Based On Image Processing

The QR code (Quick Response Code) is a two-dimensional barcode widely used as part of daily life due to its good featured such as large capacity, high reliability, all-direction high speed scan, small printout size and effective Chinese characters expression. General QR code reading software can not correctly decode QR codes due to the image problems such as uneven illumination, tilt angles and distorted surfaces.In this paper, the pretreatment technology of QR code recognition has been studied base on the fundamental principles of image processing. Three new methods are proposed to improve the accuracy of QR code decoding when dealing with the image problems such as uneven illumination, perspective distortion and curved surface distortion.Firstly, the QR code binarization algorithm in uneven lighting conditions is proposed. Sauvola's adaptive document image binarization algorithm is modified according to the features of QR code symbols and the relative parameters are given.Secondly, a distortion correction algorithm is proposed to correct the perspective distorted QR code images taken by mobile phones adaptively. At first, the QR code image is filtered. Then, the gradient-based Hough transform is modified to calculate the adaptive parameters. The eight longest lines are calculated and the contour lines of QR code symbol are detected. After that, the four vertices of QR code are obtained. At last, the inverse perspective transformation and gray level interpolation are performed to correct the distorted QR code image.Thirdly, a correction algorithm for the QR images printed on the curved objects is proposed based on the finite element shape function. The minimum bounding rectangle of the QR code symbol is used as the correction reference. With the four vertices and the four mid-points of contour lines of QR code symbol, the Pascal's triangle shape function is performed to transform the distorted image into a standard rectangular relatively. At last, image re-sampling is performed to reconstruct the QR code image into a standard one.Finally, the decoding software for the standard QR code symbol is achieved according to the C++language after the national standard decoding algorithm is improved.