Study On Quality Assessment Of Iris Image

Iris recognition as an important biometrics identification has received more and more attentions for its uniqueness,stability,reliability and no violation,and obtained the corresponding results.However,in recognition speed and accuracy,it is difficult to achieve the best of both worlds,but the appearance of the iris image quality assessment makes it possible to resolve this bottleneck.In this paper,based on the research of the iris image quality requirements, proposing an algorithm of iris image quality assessment,which includes two targets:the iris image definition,the visibility of iris.In order to improve the speed of image selection in automatic iris recognition system,above two indicators were combined to the target of "integrated quality assessment of iris image".At the same time,the algorithm of iris localization has a certain modification,whice can improve its accuracy.On the definition of iris image,according to the characteristics of image in gray value,using "Laplace(8 neighborhood differential) operator"to real time determine the definition of the input image.If the image is sentenced for the fuzzy,then further through the improved algorithm to analysis the causes of ambiguity,the algorithm is based on the Laplacian sharpening with the FFT(Fast Fourier Transform).The paper proposes a method of the visibility of iris,which is expressed with the weighting factor of iris and scaling factor of pupil.Among them,the weighting factor of iris is computing through the number of iris pixels,which is according to the characteristics of image in gray value;the other is computing through geometric approach,which uses the results of the iris image localization.On the integrated quality assessment of iris image,which is defined as: Q=ω₁Q₁+ω₂Q₂.Among them,Q₁ and Q₂ respecticively on behalf of the quality assessment factor of iris image definition and visibility,ω₁ is the weight of Q₁,ω₂ is the weight of the Q₂.The indicator of Q can reflect the changed relationship of iris image quality with Q₁ and Q₂,and the algorithm is simple,easy to implement.The algorithm and ideas that this paper present have enriched the study of iris recognition technology,and can improve the accuracy of identification,so which have a theoretical and practical significance.Nevertheless,there are still shortcomings in some areas(for example:iris image quality assessment is dealt with after the localization,no doubt,which will repeat with the work of iris recognition),so which needs to further study.