Robust Image Processing based Real-time Optical Mark Recognition System

Optical mark recognition (OMR) is the process of reading marked information automatically and is globally used in automating examination marking and other forms of marking. However, the current OMR techniques are often time-consuming, expensive and require dedicated hardware for grading. Moreover, often the OMR sheets are not scanned properly due to untrained staff, sheets being damaged or put in an incorrect orientation. Therefore, to circumvent these drawbacks of existing techniques, we propose an image processing based robust model for evaluation of OMR sheets. The method is highly accurate and can identify the responses from OMR sheets with lower time and computation complexity. Robustness of the proposed model has also been tested against angle variations, blur and light imbalance. It was found that the model can handle all these artefacts efficiently, making it suitable for real-time applications. The proposed model is also low-cost and portable, as images of OMR sheets can be clicked through mobile phones without the need of scanners.