Abstract

Optical Mark Recognition (OMR) has been used to gather input from large groups of people for many years, with applications including multiple choice exams and large scale voting. The need to use special purpose hardware to process the forms, and restrictions placed upon the layout of the form so the hardware can process it, means OMR is not suitable for small-scale operations such as a test for a single class of students. The data collector ends up reverting to manually processing the forms. Can this problem be solved using simpler hardware, such as a scanner or a webcam?

This talk will briefly cover some standard machine vision techniques for OMR , and explore some of the subtleties that exist in different approaches.