Invoice Image Processing System Based On Android System

Invoice is a proof of purchase,it can play an important role no matter in people's live or work.The most common and simplest way of invoice information extracting is to identify manually.However,manual identification has its shortcomings,its recognition is slow,identifier prone to fatigue,false positives,and so on.So with the social development,invoices are becoming more and more widely,and manual identification speed has been unable to meet the current requirements of intelligent and automated processing.Therefore,automatic invoices image processing need a solution to solve the problem of recognition efficiency.Invoice image processing is an image processing method,it treat invoice images as an object of analysis processing,the purpose of which is to extract the useful character information.Invoice image processing system has its own characteristics,which is mainly reflected in the invoice image quality,it could not get the desired results when using common image processing method for processing.To this end,this paper presents an invoice image processing method for low-quality invoice images,to solve the efficiency problem when using other image processing methods for low-quality images processing.Invoice image processing algorithms presented in this thesis consist of three main steps,namely image binarization,region segmentation,information recognition.For each part of the algorithm,the principles are analyzed and the detailed steps are designed.Binarization is the core of invoice image processing algorithm.Binarization method presented in this paper is based on gray models of images.Based on this model,binary threshold formulas and procedures are designed.Experimental results show that the proposed image processing algorithms can be running on a variety of relatively poor quality equipment to shoot and get the desired results.Another important design in this paper is the design and optimization of the content of the invoice image processing system.For the difference between algorithms and interfaces,we designed two different design methods and procedures.Use OpenCV which runs in Native mode for programming of algorithm,and use multi-threaded technology for the design of UI.Finally,this paper analyzes the program of algorithm,proposes an optimization scheme and got very good results.