A new QR Code Recognition Method using Deblurring and Modified Local Adaptive Thresholding Techniques

Quick Response (QR) Code would be easily affected by motion blur. In this study, a new recognition method for blurry QR Code is presented. First, a motion deblurring algorithm based on generative adversarial networks (GAN) is applied to perform the blind motion deblurring on single QR Code image for improving the quality of QR Code pictures. Then a modified local adaptive thresholding algorithm applying integral image technology is utilized in order to acquire binary QR Code pictures. Finally, this algorithm would be compared with some previous methods. Furthermore, for testing this algorithm, a database which contains motion blurry QR Code images from both public database and real scenes is collected. Experimental results show that this proposed algorithm perform well on motion blurry QR Code recognition missions.