Open AccessNon‐ideal iris segmentation using anisotropic diffusion
Author(s) -
Wan HongLin,
Li ZhiCheng,
Qiao JianPing,
Li BaoSheng
Publication year - 2013
Publication title -
iet image processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.401
H-Index - 45
eISSN - 1751-9667
pISSN - 1751-9659
DOI - 10.1049/iet-ipr.2012.0084
Subject(s) - iris (biosensor) , segmentation , artificial intelligence , computer science , pyramid (geometry) , iris recognition , anisotropic diffusion , diffusion , boundary (topology) , computer vision , image segmentation , pattern recognition (psychology) , anisotropy , ideal (ethics) , mathematics , image (mathematics) , geometry , optics , mathematical analysis , physics , biometrics , thermodynamics , philosophy , epistemology
Iris segmentation is critical for iris recognition. In this study, the authors present a circle‐based iris segmentation method for non‐ideally captured iris by employing anisotropic diffusion. Our proposal consists of two component steps by which interior and exterior boundaries are localised, respectively. To save computational load, Laplace pyramid (LP) framework is incorporated into both steps. During the first step, when iris has been decomposed into the coarse level by LP, reflections will be removed by anisotropic diffusion and morphologic operations. In the second step, the authors present the innovated curve evolution to detect exterior boundary. Moreover, order statistical filters are employed to enhance the contrast of iris and sclera. Experimental results depict a high correct ratio of segmentation that is more than 96.90% thereby validating the proposed approach.