Research On The Automatic Recognition Techniques For QR Barcode In Blurry Images

Quick response(QR)barcode is a common two-dimensional barcode.Compared with other 2D barcodes,QR barcode has the comprehensive advantages such as high encoding density,fast recognition speed and strong capability of chinese character encoding.This makes its wide applications in our country.However,QR barcode image is usually captured by the cameras of mobile phones.Its capture is constrained by various factors such as the spatial resolution of mobile camera and the lighting conditions.Thus,the captured QR barcode images might suffer from bluriness,uneven illumination,geoemtric distortion and even partial damages.These complex conditions bring side effect to the automatic recognition of QR barcode,which requires advanced image pre-processing techniques to improve its recognition ratios.Therefore,the automatic QR barcode recognition under complex conditions is a hot research topic in the field of barcode applications.To address the issue of automatic recognition of blurry QR barcode images,this thesis researches on the image pre-processing techniques involved in it.Specifically,the main works are summarized as follows:First,the image pre-processing techniques for the automatic recognition of QR barcode in blurry barcode images are discussed in details.It include the key steps such as binarization,rough location,fine location,rectification of geometric distortion and adaptive sampling.Specifically,an efficient iterative deblurring approach is proposed on the basis of Constrained Least Squares filter.It combines the incremental Wiener filter and object-level constrains to improve the visual quality after the image restoration of QR barcode.For the binarization of QR barcode,a novel local thresold selection approach is adopted based on the Sauvola's approach.Moreover,an adaptive grid sampling approach is presented,which adpatively divides the regions and samples the grids by combining the Finder patterns and Alignment patterns.Second,the modular programming design is adopted,and a complete automatic recongition software is programmed with C# for QR barcode images.On the basic of standard QR barcode decoding,the flowchart of automatic recognition for QR barcode images captured by mobile cameras is presented.And the divisions of its software modules are also given.Practical examples of several QR barcode image recognition show that the proposed image-processing approach can correctly decode the QR barcode images captured by oridinary USB cameras,even when they suffer from complex distortions.No matter these barcodes are printed on paper or displayed on computer screen.Moroever,the recoginition speed is moreefficient than Zxing.The proposed image pre-processing techniques for QR barcode images can achieve relatively high recognition accuracy and speed,which proves the desirable recogntion performances.