Survey evaluation by character recognition

Abstract

In the study, an automated system was proposed for the evaluation of survey sheets filled by different students. In this method, the regions related to the answers on the survey sheets digitized by a scanner are determined. For this purpose, after finding the right edge of the survey, upper-right corner of the survey is marked by a developed edge tracking algorithm. Afterwards, rows of the survey are found and the cells which contain the answers between the rows are segmented. After the pre-processing step that includes filling and thinning operations, the answers in these parts are categorized using histogram-based features and area ratio. Obtained performances using limited survey sheets indicate that the survey evaluation by character recognition can be an appropriate option.