Font Size: a A A

A Fast And Effective Iris Segmentation Algorithm In Complex Environment

Posted on:2016-10-12Degree:MasterType:Thesis
Country:ChinaCandidate:Y H LinFull Text:PDF
GTID:2428330461976155Subject:computer science and Technology
Abstract/Summary:PDF Full Text Request
Iris segmentation aims at isolating the effective iris pixels from original image for similarity matching in iris biometrics system.A reasonable segmentation algorithm includes reflection removal,pupillary and limbic boundary localization and sometimes eyelids and eyelash detection.Iris segmentation is considered as a forepart module and plays a key role in iris biometrics.However,under non cooperative or low cooperative iris acquisition system,it often appears strong reflection,serious occlusion,hair noise and off-angle aberration.Compared to traditional and popular methods,this paper afresh the mathematical model definition of process objects through more associate with physiological characteristics and presents a fast and effective iris segmentation algorithm.First we adopt a signed integro-differential operator and ring-shaped model to reflection removal.Follow,the region of interested is extracted by the adaboost classifier based on haar features.Then transform the region into polar coordinates and gray gradient operator performs for edge and boundary enhancement.Finally,the proposed spindle ternary tree model is used to track iris boundaries by searching the best route.Under the proposed model and algorithm,partial noises influences would be weaken and smoothness of the boundaries could be ensured.A similar tree also can be adopted for eyelids localization after the rank filter operation under the cartesian coordinate system.Apply this algorithm in iris recognition system,compare the experimental results with current popular algorithms,this algorithm performs better performance in some evaluation indexes contains recognition accuracy rate,false accept rate,false reject rate and equal error rate.
Keywords/Search Tags:Biometrics, Iris Segmentation, Reflection Removal, Gray Gradient, Ternary Tree, Eyelid Localization
PDF Full Text Request
Related items