Image Restoration Of Two-dimensional Barcodes Based On Regularization

With the rapid development of the information technology, the two-dimensional barcodetechnology is used on varieties areas. However, in the case of complex environmental background, thetwo-dimension barcode decoding will be influenced by some bad conditions. All these degradationson barcode image are not conducive to the recognition and decoding of barcode. The thesis proposes anew algorithm with regularization. The algorithm uses the prior information of local variance ofbarcode edge area to construct a new regularization item and also construct its reasonable parameterwith restoration image of each iteration based on three degraded type: motion blur image, defocusblur image and Gaussian blur image, it solves the problem of different degradation types’two-dimensional barcode image recovery. It will evaluate and analysis the restored algorithm by usingpeak signal-to-noise ratio and contour-volume and then study its distributed of the gray color oftwo-dimensional barcode image itself. And at last, it designs a system using that algorithm to recoverythe two-dimensional barcode from the container equipment and gives the recognition result. The mainwork is divided into the following sections:First of all, the paper introduces QR barcode structure information and code system, themathematical model of the three common degradation types and the evaluation of image restorationmethods of no-reference and full reference.Secondly, the paper proposes a new algorithm with regularization technology that uses the priorinformation of local variance of barcode edge area to construct a new regularization item, adding theedge regular item to the end of the mathematical model of normal regularization method. Thatalgorithm refines the edge area of two-dimension barcode on the restoration of smoothing constraintitems to achieve the optimization of barcode image restoration effect.Thirdly, the paper constructs the parameter of the edge regularization item and analyses thequality of that algorithm with the non-reference and full reference image restoration evaluationmethods, together with the gray pixels distribution of two-dimension barcode.Fourthly, the paper designs a system using that algorithm to recovery the two-dimensionalbarcode image and gives the recognition result by capturing the QR image from web camera and theMatlab and C mixed programing platform.The proposed method is based on edge regular item of recovery, which solving thetwo-dimensional barcode image with gray-scale pixel distribution. It has a better quality of recovery.It can meet the requirements of the two-dimension barcode in the practical application.