Fast Identification Of Low Quality QR Code And Its Software Design

QR code has several benefits, including strong capacity for information storage, easy to be read, the pretty good privacy, etc., which has been widely used in the documents management, the government, the public sector, the circulation of commodities, electronic tickets, train tickets, website chain and many other fields.Traceability system of fruit can effectively reveal and transfer information by using QR code. So, traceability system is gradually used in farm production quality control systems,which can enhance its safety. And consumers can obtain the corresponding detailed information by scanning QR code, such as the original operation, growth and harvest, the fruit quality and circulation, etc. And it can strengthen the fruit market competitiveness. But because QR code is susceptible to complex background, noise, uneven illumination and the influence of distortion, which cause the quality decrease sharply of QR code image and make it difficult to realize and recognition. For eliminating these effects and improving the quality of low QR code identification efficiency, based on image processing, I studied the portable terminal of the low quality QR codes quickly identify, and developed smart-phones QR code identification software. This paper mainly studies QR code identification methods, including image processing algorithms, such as, noise immunization algorithm, uneven illumination elimination algorithm, position, and distortion correction algorithm, and based on portable terminal Android cellphone platform software design, etc. Specific work contents and concludes as follows:(1) QR code image with complex background is difficult to quickly extract, so I researched the visual attention mechanism, including Itti and GBVS models, respectively extraction QR code image under complex background, compared their extraction effect. Tests show Itti model can more effectively eliminate complex background and extract the comprehensive QR code image, so this article choose Itti model to divide QR code in complex background.(2) QR code application exist noise condition, so I studied many kinds of noises, such as additive and multiplicative noises. Then I researched QR code image noise immunization methods, such as mean filter algorithm, median filter algorithm and wiener filter algorithm.The results show the median filtering algorithm has the highest average PSNR 41.07 to dealwith the noise results, wiener filtering denoising processing results with a 19.34 average PSNR, and average filtering denoising processing result average PSNR centered at 26.41.So this article selects median filtering to eliminate the noise.(3) QR code is easy to be influenced by uneven illumination problem, so I studied a variety of binarization algorithms to eliminate the influence of uneven illumination, by comparing the Otsu method, Retinex method, histogram equalization method and the theory of illumination invariant image in QR code image processing effect, and selected an optimal eliminate uneven illumination algorithm. The results show that the Otsu method after16 partitioned histogram equalization is the best, its recognition rate is 94.4%, higher than Retinex image enhancement algorithm and Otsu algorithm respectively 43.3% and 10%, can get rid of the influence of uneven illumination.(4) For QR code rapid positioning problem, I studied the fast localization algorithm, such as position detection image and the edge of the localization algorithm based on Hough transform. Due to the large amount of calculation, Hough transform on the mobile terminal can’t well realize rapid, real-time and efficient requirements. So, I selected the QR code structural characteristics of location detecting graphics to locate.(5) The spatial relationship between pixels of QR code image is changed during the processing of the obtaining image, which is prone to geometric distortion problem, so I studied geometric distortion correction algorithm for QR code, including the space transform and gray level interpolation. Geometric distortion correction is to correct distorted QR code in the image pixel position to regain space of original relationship between pixels. At the same time, gray level correction is to restore the values of original pixel. Through space transform and gray level interpolation, the distortion QR code image will be corrected.(6) Based on a portable terminal Android intelligence platform, I developed software for low quality QR code fast and efficient recognition, which can eliminate noise impact, uneven light, distortion, and realize the destination of fast and efficient recognition of QR code.