Design And Implementation Of OCR Card Address Recognition Post Processing System

Optical Character Recognition(OCR)is a technology that enables you to convert the text image into editable data through the processing and analysis of the text image.Therefore,the technology as a convenient,simple character input method is very popularity in this age of huge quantities of data and information.With the popularity of the Android mobile phone platform and the continuous progress of information,combination of OCR card address recognition and Android mobile device has been observed more and more applications and popular in the intelligent Electronic Commerce Management.However,in the process of OCR identification,recognition rate is affected because of that the structure of the text and the image itself is complex and changeable,besides image sharpness is also due to the performance of the device image acquisition.Card address information containing state,provinces,cities or villages,streets,doorplate number,housing estates,buildings and so on.An effective address should be unique,to help business visit,job search,company registration,etc.The analysis of a large number of actual business cards address reveals that the actual business card address information exists some address information description is not complete,which is not conducive to intelligent management and application.To solve the above problems,this paper studies the OCR card address post processing system,in order to improve the recognition rate and provide more standardized address,and make card management and application more intelligent.In this paper,the design OCR card address recognition post processing system based on the Android platform via a cell phone camera shot paper document in image form entry phone.And combination of Android NDK programming and OCR post processing to obtain the input image corresponding address which is accurate and perfect output.The whole system is divided into three parts:?Image acquisition process.The system uses Android mobile phone camera to take pictures and collect the business card.?OCR business card address identification process.This part is not as the focus of this article to explain the content.The system through the JNI technology integrated OCR engine to complete card address recognition.?The last part is the OCR business card address recognition post process.This part is also the focus of this article.And the post process is divided into three steps: First step,word segmentation processing for engine initial recognition address.Word segmentation uses forwards maximum fuzzy match algorithm based on dictionary.Second step,querying these addresses of dictionary matching in database.Meanwhile querying associated address information.The last step,original address correction and filter the final result by combining the similarity calculation method which is based on edit distance and Chinese address rule.Experimental results show that the system not only improves the accuracy of OCR recognition but also perfect address information itself intelligently,as to get more effective address information.Experiment shows that,the system improves error rate partly but improves perfect rate obviously.