Research On Distortion Correction Method Of QR Code Image Based On Adaptive Anti-aliasing Mechanism

With the rapid development of the Internet,the two-dimensional code has been widely used in fields such as identity verification,logistics traceability,and mobile payment due to its advantages in terms of storing large amount of information and rapid decoding.However,during the acquisition process,the accuracy is easily affected by the imaging environment and two-dimensional code geometric distortion also make negative influence to the decoding process.Therefore,correction of the two-dimensional code is very critical.During the process of correcting the distortion image,it is necessary to resample the image.This process is prone to aliasing,where the corrected image appears new deformation and still hard to be recognized.Therefore,it is of great significance to study how to reduce aliasing during correcting process.The standard Gaussian convolution kernel is usually used for anti-aliasing processing.Although this method can suppress some high-frequency signals,it will cause post-aliasing.In addition,how to select parameters correctly is also particularly critical for the correction effect.If the parameters are incorrect,new deformations will occur in result which would also affect post process.Based on the above problems,this thesis studied the anti-aliasing mechanism in the process of distortion correction of the planar and cylindrical QR code images.The main content and innovations of this thesis include:(1)Aimed at selecting the optimal homographic matrix in the correction process,this thesis proposed a homography matrix estimation algorithm based on the feature points from two-dimensional code image,which detects the image vertex and makes similarity judgment to make the distortion image to a standard template.To work out the aliasing problem of traditional methods,a homographic model correction algorithm based on mapping adaptive convolution has been proposed,which can generate high-quality corrected images and boost the recognition rate.(2)To select the cylindrical radius parameter of the cylindrical QR code image,this thesis proposed a cylindrical model parameter estimation algorithm based on the dichotomy.Through this method,the radius of the cylindrical surface can be accurately acquired,so that the corrected image is a standard two-dimensional code.In order to resolve the pre-aliasing and post-aliasing problems caused by correction process,a QR code correction algorithm based on cylindrical mapping has been proposed,which can effectively address the aliasing issue and improve the recognition rate of cylindrical QR code.(3)To tackle the aliasing problem of QR code images with compound distortion,this thesis proposed a QR code anti-aliasing correction algorithm upon compound mapping.A composite map is constructed by mathematical methods and the deformation convolution kernel is derived.This method can effectively fix the aliasing problem of the two-dimensional code image with both perspective distortion and cylindrical distortion during the correction process,and increasing the recognition rate of composite-distorted image.