Optical signal recognition system using image processing techniques

In this thesis, it is aimed to develop Optical mark Recognition (OMR) system with image processing technology. In our country and in the world education systems, optical forms with multiple choice choices are frequently used both in the evaluation of students' success and in the selection of students. Optical reading devices of optical forms are evaluated by using optical mark recognition techniques. In this thesis, it is aimed to evaluate the results obtained with a system design and application for the reading of multiple choice test exams answer papers based on visual processing. The system was developed using Microsoft Visual Studio 2013 with Visual Basic (VB) programming language. Reading and evaluation of exam papers is an important activity which takes a significant part of the time of the trainers. Accurate and accurate operation of this process is of utmost importance for the evaluation of education. In addition, this system will be able to process quickly for processing hundreds or thousands of optical answer sheets. One or more options are identified in the system. The method used is based on the calculation of template and the key point. The images that were created after the OMR answer sheets and the answer key template were transferred to the computer with the help of the scanner were used as input. This system works in four stages. In the first step, the template is created by selecting the coordinates (the working area of the starting point and the end point, the length and width of the starting point and end point, the size of the marking area and group regions). All coordinates are saved. As the second stage, the key point detection algorithm is applied. As a third step, the optical answer pages are automatically divided into three areas of interest (student ID, student name and multiple choice questions) and calculate each bubble column or black pixel in row using projection profile and threshold technique. The projection profile and thresholds are used to define N-row horizontal lines and N-column vertical lines. Horizontal lines can be applied to determine multiple choice questions, and then can be classified into the threshold (using Bernsen's Technique). At the same time, the vertical line can be applied to identify the student ID and student name, and then be classified into the threshold. In order to determine the character limit in the literature, he applied horizontal and vertical projection profile in the majority of scientific studies. In this thesis, projection profile was applied to determine a bubble limit. In the fourth stage, a statistical result is obtained for each student by comparing the exam paper with the answer key in the previously recorded file and automatically calculating the number of correct answers. In this thesis, more than 100 exam papers were tested. As a result, it was realized with a 100% accuracy rate for the area determination consisting of five different formats of multiple choice answer sheets. The processing time for each paper of the system is less than 1 second. The performance of the developed system is compared with three different studies. These comparisons focus on the performance comparison and accuracy rate of processing times. All test forms used in the comparison process contain between 25 and 100 questions. The result of the comparison procedure took more than 1 second for each examination paper. In addition, the results of the study were performed with an accuracy rate of 97.6% - 100% depending on the study.