A comparative study on hough transform segmentation approach for outlier iris image

Abstract

Non-ideal eye image is a contemporary issue that was attracted many researchers to contribute a solution in iris recognition domain. It consists of iris occlusion, motion blurred, off-angled, and outlier’s situation issues. However it is rarely discussed on outlier’s situation in recognition field related with the ophthalmology domain. Therefore this study was motivated by the outlier’s situation issue in order to prepare a reliable ophthalmology system by using Hough transform segmentation approach. Three frequently used iris databases will be applied in this study to test the segmentation approach and they are CASIA, UBIRIS and UPOL. The result from the conducted testing is to compare and evaluate the accuracy of the approach to the databases. Hence, the consequence from this study is to emphasize the issue that influence the accuracy of the segmentation process.