Study On Recognition Algorithm Of QR Code Based On The Mobile Phone Camera To Scan

With the development of information technology, two-dimensional codetechnology is applied more and more widely in our life. QR code, as a very good2Dbarcode in the world, also spreads rapidly in china at present. Since one-dimensionalbar code can no longer meet the needs of social development, the two-dimensionalcode technology received more and more attention. As the amount of informationstored in the one-dimensional bar code is limited, one-dimensional bar code can onlyidentify a single object, not including detailed description of goods. The developmentof modern information technology and the needs of representing large amounts ofchanging information increased the desire for barcode to express more information inthe limited geometric space.In recent years, mobile phone’s function is more and more powerful and themobile communication business is also development unprecedented. With the rise ofMicro message M chat and micro-blog communication software, the application oftwo-dimensional code also formed a huge user group network. Therefore, theinformation identification of the two-dimensional code on mobile phone has become ahot research issue. At present, some QR code recognition software can read the generalcode of two-dimensional image well. However, to those two-dimension codes whichare seriously distorted, partially losing or printed on the surface easy to wrinkle, theseQR code recognition software have a very low recognition rate and even cannotrecognize them. Therefore, it is necessary to study the pretreatment of thetwo-dimensional code image before reading.This thesis mainly did the pre-processing work for the two-dimensional codeswhich are distorted seriously and the existing two-dimensional code image softwarecannot identify. In order to eliminate the influence of salt and pepper noise of QRimage, an improved adaptive median filtering algorithm is proposed based on theclassical adaptive median filter. The experiment results show that the improvedalgorithm is more suitable for processing QR code image than the original one. For theproblem of image pretreatment of two-dimensional code images which are notdistributed uniformly in the light or occur geometric deformation when shootingbarcode, this thesis put forward a method of correction based on BP neural network.The new method used adaptive neural network to complete correction of image distortion by fitting the geometric deformable model. The experiment results show thatthe algorithm has a certain effect in improving the rate of the two-dimensional codeimage recognition.