Iris Segmentation with Using Hyperspektral Images

ÇİNAR L., Tasdogen A.

24th Signal Processing and Communication Application Conference (SIU), Zonguldak, Turkey, 16 - 19 May 2016, pp.1093-1096 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume:
  • Doi Number: 10.1109/siu.2016.7495934
  • City: Zonguldak
  • Country: Turkey
  • Page Numbers: pp.1093-1096
  • Hacettepe University Affiliated: Yes


The aim of this work is to improve the performance of iris recognition systems by using hyperspectral iris images. The commercial iris recognition systems that already exist mostly use the information of iris in the infrared band and the performance of these systems more often depends on segmentation process. Thus, in this work, more reliable and effective segmentation is attempted by utilizing iris information in visible and infrared bands. The dataset that includes iris images of 17 people in 16 different wavelengths is used. The circular Hough transform is used to detect the region of the iris. K-nearest neighbors (K-NN) algorithm is preferred to use the hyperspectral data information in different bands. As a result of these methods, segmentation is made with 84% success rate.