Single cell gel electrophoresis, also known as comet assay, has been widely used for assessing the effect of genotoxicity and detecting DNA damage of individual eukaryotic cells. There exist established imaging techniques for cometassay analysis, but these platforms have limitations such as required user interventions, low throughput, and weakness to noise caused by incomplete dyeing of fluorescent materials and other experimental errors. To resolve these, we propose a novel procedure for analyzing comet assay images, which considers various DNA damage patterns and classifies them in a robust manner. We tested our approach with twenty golden data sets containing over 300 comets and achieved satisfactory classification accuracy.