Study On Key Techniques Of Marking System For Objective Questions And Figure Correction Questions In Basic Mechanical Courses

Exam marking is indispensable in the education industry.Owing to the needs for informationalized marking,new marking methods has emerged and widely been used,such as paperless marking system and automated marking system.Taking account into the actual marking requirements of mechanical courses,the key techniques of automatic marking for objective questions and figure correction questions has been studied in this paper.And the system of marking objective questions and figure correction questions has been developed,which is significant to reduce teachers’ burden and improve the teaching quality,examination reform and curriculum evaluation system.First of all,the architecture of objective questions and figure correction questions marking system has been given.C/S architecture model of the system has been analysed.And according to the operation process of system,database has been designed.Secondly,according to the system requirements,key technologies of objective questions scanning and marking system are analysed,including template customization method,image preprocessing and image clustering method based on K-means improved algorithm.The abstracting and modeling for the answer sheet with answer template have been given to achieve abstract description of any kinds of answer sheet and to avoid the dependence on specific structure of answer sheet.Different skew detection and correction algorithms have been given for different types of answer sheets,so that the system can be compatible with various types of answer sheets.Improved K-means algorithm has been applied to the cluster recognition of answer box for objective questions.The local cluster density has been defined to avoid selecting the initial cluster center randomly,and clustering result has been evaluated by discrimination.The test results have shown that stable clustering results can be obtained by the improved K-means algorithm.And frequency of iterations can be reduced and recognition accuracy can be improved.Thirdly,according to the requirements of automatic marking of mechanical courses,key techniques have been analyzed,such as image lightening,attribute setting and answer discriminating.Influence of original lines on examinee’s answer is reduced by binarization,image refinement and image grayscale adjustment.Standard answers are divided into four types and setting principle and discriminant principle have been given by the summary and analysis of various figure correction questions.Finally,with application examples the marking system for objective questions and figure correction questions has been tested.The test results have shown that stable clustering results can be obtained by the improved K-means algorithm.And frequency of iterations can be reduced and recognition accuracy can be improved.With the assessent indexes of test questions,paper analysis and score analysis,informationization of mechanical course examination reform can be achieved effectively.