مرکز منطقه ای اطلاع رساني علوم و فناوري - Reading Optical Form From Web Camera With C #

Abstract :

One of the most important difficulties of the test methods that are often used to evaluate the students is to read optical forms without mistake and to obtain necessary statistical information. In this study, to make use of image processing techniques widely used in every area, especially in education area is aimed. A programme which can read optical forms with the web camera using C#.NET is developed. Firstly a test mechanism is designed. A standard 1.3 megapixels, branded Philips SPC 900NC, charge coupled device camera is fixed to this mechanism. At the same time, with the help of fixture, optical forms are ensured to be placed into the test mechanism in the same direction. The image which is 640×480 pixels and in RGB format is firstly transformed into grey level then the image is concretized by stretching contrast. According to dynamic verge rates obtained with Otsu algorithm, transformation from grey level to binary level is made. The coordinates of the signs which are on the left and right top corners of the form are determined. According to these signs, the central coordinates of all the options in the form are figured out. With the developed algorithm, total of the black pixels is evaluated, signed options are fixed and registered to the database. Signed options are compared with the answer key and the number of the answers which are true, false, empty or signed wrongly is obtained. Statistics of the exam analysis is gotten by evaluating all the answer sheets belonging to the exam. At the end of the study, the area in the optical form for coding the students numbers and the type ofthe booklet is read without any mistake and %100 percent of success is gained. The average of the success of the areas including test1 and test2 is %99.90. In other sides, only 7 of 221 optical forms is read inaccurately, 214 optical forms are read without mistake and %96.83 percent of accuracy rate is obtained.