Study On Automatic Identification Of Address Information In Express List Based On Machine Vision

With the internal informatization of the express company,the delivery documents are in the stage of the transition of ordinary documents to electronic documents.At present domestic most of the small and medium-sized express company edcs standardization,informationization level is low,still rely on manual entry address information,to express business to realize automatic sorting brought obstacles.This article from the perspective of system hardware and software.This paper studies the automatic identification of address information in two types of express list from hardware and software,so as to provide favorable support for the unified management of the express address address information and optimization of the sorting business.Firstly,according to Nyquist theorem,this paper analyzes that the image sampling accuracy needed for express list address text and bar code needs 0.065 mm,and according to the precision requirement,an automatic recognition system based on automatic zoom lens is built to realize the the image of express list with a height range of 1500 mm.The test indicates that the time required for complete processing of a parcel is 1.1s,and the dimension error of the field of view is less than 1.0%,and the repeated stability is good and the image quality meets the system requirements.After obtaining the imaging system information,the upper computer first processes the binary,positioning,and extraction of the express delivery image.For ordinary list,it is necessary to recognize the handwritten address characters and barcodes.This paper proposes a method based on the combination of address database and BP neural network to realize the recognition of the handwritten address characters.For electronic list,barcode recognition is achieved by calculating the width of black and white bars.The experimental results show that the accuracy rate of the automatic recognition system for the ordinary list is 92.7%.In the recognition of electronic list,the accuracy rate can reach 99.4%.