Paper Roll Digital Image Recognition In The Online Marking

The speed of computer process and operation are becoming faster and faster, with the rapidly development of computer technology and information. Because of those factors, the applications of Online Marking System (OMS) are being spread and applied rapidly. OMS is based on network technology and electronic scanning technology, with principle of high rate of accuracy and realization of fairness in the examination.Now, there are subjective questions and objective questions in all kinds of examinations. So, OMS are composed of paper digitization system and online subjective questions marking system. Paper digitization system is used to marking objective questions automatically, and it must have high rate of accuracy. The purpose of online subjective questions marking system is to realize the justness and fairness in the examination.This thesis is focused on automatic marking system in the paper digitization.Firstly, current situation of OMS and theory of image process are introduced. These technologies include the process of paper digitization, image grayscale and some methods used to eliminate noise.Secondly, during examination papers scanned, some papers will be rotated by human factor or the quality of scanner. Those factors will have great impact on rate of marking accuracy. Therefore, how to find a method to solve those problems is very important. In this thesis, we used some special blobs in the examination papers as reference points to rectify image and to find the most optimum combination of reference points at the same time.Thirdly, the most important process of objective questions (OMR) recognition is binarization. According to the feature of paper's qualifications will be impacted by surrounding environment, we offer a new method to get self-adaptive threshold value and describes in detail the theory and technology of this method.Finally, the theory and method of OMR recognition is described in this thesis. Based on the demand of fast speed and high accuracy rate of recognition, a way that use paper's template and fuzzy search technology to rapid location and recognition is introduced. Ultimately,the rate of OMR recognition is accurate.