Improved and robust eyelash and eyelid location method
Ting Wang, Min Han, Honglin Wan · 2012
Iris recognition has been very popular among researchers as an important personal identification technology due to its unique, stable and noninvasive properties. However, because of iris occlusion such as eyelid and eyelashes, high accuracy of iris recognition system is challenged. In this paper, we firstly improve our previous work on eyelashes localization algorithm based on Expectation Maximization (EM) and Gaussian Mixture Model (GMM). Then, we propose a novel and robust approach to search the eyelid via hybrid edge detection and Hough transform, which reduces the noise fitting points and selects the eyelid fitting area automatically. Experimental results reveal our proposal can detect eyelid and eyelashes accurately and effectively.