Design Of Test Paper Automatic Correction System Based On Machine Vision

In the current education and teaching,Examination plays an important role,But there's lots flaws in current examination paper correction.The first flaw is duplicate work and Inefficiency.The way of manually correcting test papers will waste lots time and energy.Second,there are many disadvantages in the way of using answer cards,The answer cards paper is a specially designed and optimized paper,So the cost of answer cards is very high.And a special card reading equipment is needed,increased cost of use,unable to realize the generalization of examination.And in reality,In order to test students' mastery of knowledge,teachers often do some random classroom tests,which can't achieve by using answer card.In addition to that,we need to design different answer cards for all tests,and that is a waste of energy and time.For students,the answer card requires a special pencil,it takes time and effort to revise the answer,it may be occur incorrect identification because of the imprecise alteration,it is more suitable for large-scale examination rather than smaller-scale examination.There're so many defects of the answer cards,so we should design a low-cost,easy to use and universal correcting method.The thesis designs an examination paper correcting system based on machine vision,applying machine vision to the examination paper correcting,to achieve the success of automatic marking and correcting test paper or aid correcting.This system can be used to correcting papers instead of or help teachers,improve the working efficiency in educate.For the design requirements of test papers automatic correcting,this thesis is focus on:image processing and character recognition algorithms,through digital image processing,We extract the answer information from the test paper,Convert the answer character into a binary image.Character feature extraction method.extract the most distinguishing features of different characters from the normalized binary image.inorder to distinguish different character.Character recognize.Through machine learning we get an input-output model,and get the characters by machine learning.This system use OpenCV+VS2013 and based on computer platform to achieve automatic correcting of test papers.The main innovations of the thesis are:To solve the defects of using fixed answer box,This article designs an interactive answer box,the user just need to draw a rectangle box of each question where the answer may write,and it can be used in every kinds of test pepper with good compatibility.Assist in score statistics,Analyze students' scores,generate visual data,help teachers analyze students' mastery of knowledge points.Through the test,this system shows a good effect at automatic grading of test papers and it can meet the basic use requirements in education and teaching.