Based On The Portable Devices Two Dimensional Barcode Identification

With the growing popularity of two-dimensional barcode applications, barcode readers usher new opportunities. Two-dimensional barcode devices are inclined to mobile, portable and economic developing gradually.Compared to the fixed devices,the problem of barcode recogniton by portable devices is the variable environment of barcode,such as the shooting distance,shooting angle,or twisted shooting and so on. In addition,the CPU and memory of portable device is much smaller than the PC,so the time complexity and space complexity of barcode recognition algorithms should be limited within a certain range.This article is to recognize the two-dimensional barcode by portable devices efficiently,especially about the distorted barcode image.Before the identification, barcode image preprocessing is needed, including grayscale processing, histogram-based median filtering and block OTSU binarization.The next is to filter out the text and other patterns in the image and locate the barcode. In this thesis an algorithm that recongnizes connected region from barcode image by marking pixels is implemented.Geometric correction to the distorted barcode image should be done after getting the barcode area. In this paper, from comparison of the three correction methods,such as Coons method, the finite element method,and LMA-based perspective transformation,we can see that Coons method and LMA-based perspective transformation can get satisfactory result. To the nonlinear deformation of the barcode image, an algorithm is presented in this thesis that identities module based on polynomial fitting.In this algorithm the geometric correction should be done firstly,and then fitting the row and colomn edge of each moudule by least squares method,the last step is recognition module.Compared to the existing two-dimensional barcode recognition algorithm,this algorithm can overcome the impact of nonlinear deformation better, and significantly improve the recognition result.