Research On Fast Blind Deblurring Algorithm For QR Code

With the development of the Internet of Things technology and information technology,the Internet provides an environment where people can share information without being limited by time and space.As the main entrance of the Internet,the two-dimensional barcode,especially the QR code,has the characteristics of fast recognition speed,little space occupied,large data density,and wide application range,which is widely used in various fields.However,in the imaging process of systems such as express sorting,the parcels with QR codes are sorted.When the lens shakes,looses,falls,or optical noise and so on,it would cause the image out-of-focus blurring,which makes the sorting device unable to obtain the internal information of QR code,resulting in low sorting efficiency and unnecessary sorting cost.This is a common phenomenon of image degradation.In addition,in the era of commercialization,especially driven by 5G technology,people have higher requirements for information acquisition speed.Therefore,the fast deblurring of QR code has become a research hotspot at this stage.The paper mainly focuses on the problem of out-of-focus blurring of QR code images.The research contents and innovations are as follows:(1)Aiming at the slow speed of the blind deblurring algorithm caused by multi-scale iteration,a fast blind deblurring algorithm for QR code images based on adaptive scale control is proposed.The algorithm introduces the gradient function value as the evaluation mechanism on the existing blind deblurring algorithm.When the latent image can be recognized,it will be output in advance and end the iteration to achieve the effect of adaptive scale control,so as to greatly improve the speed of image blind deblurring.Moreover,a scale control parameter is introduced to comprehensively coordinate the relationship between deblurring time and recognition rate.And different parameter values can be selected according to different demands.The evaluation boundary values of QR code images from version 1 to version 5 are counted.Within the range of appropriate scale control parameters,the deblurring time is distributed from 2 seconds to 6 seconds.Compared with other algorithms,the deblurring speed is improved by one order of magnitude,and the recognition rate is guaranteed to be more than 94%.(2)Aiming at the problem that the existing blind image deblurring algorithms based on prior which do not fully consider the characteristics of the input image,a fast blind deblurring algorithm of QR code based on edge prior information is proposed.Based on the principles of step edge characteristics,optical information theory,optical imaging mechanism,blur invariants and confusion spot centroid invariance,combined with edge prior information,the algorithm can effectively obtain the value of point spread function parameter,i.e.blur radius,so as to restore the image.Among them,the edge prior information is the average distance from the edge of the clear QR code images of the same batch to the centroid.In the deblurring of natural image,compared with the existing algorithms,the average deblurring time of the proposed algorithm is 0.1396 seconds and the recognition rate is 88.59%,which has good stability and robustness.(3)Aiming at the problem of the algorithm in research 2 which could rely on the prior information of the image,a fast blind deblurring method based on imaging mechanism is proposed.This method designs a QR code image identification with anti-blur characteristics,and quickly restores the blurred image identification based on the imaging mechanism such as the invariance of blurred centroid.The identification is composed of three circular image finder graphics and QR code.The connection of the centroid can determine the position information of QR code,and the deblurring effect will not be affected by the change of the size and direction of the image identification.Compared with the other algorithm,the proposed algorithm has improved in structural similarity and peak signal-to-noise ratio,especially in the deblurring speed,and its average deblurring time is 0.3292s.