IRIS Template Classification Using Selective Sub Bands of Wavelets

Considering multiple biometric templates per user account by biometric authentication systems for high acceptance rate leads to large storage space and computational overheads. Classification of these templates into significant sub groups will reduce the above overheads. Iris templates carry very distinctive texture information such as brightness, shape, size, uniformity, directionality, regularity etc .Iris texture classification based on wavelet pattern analysis is one of the most effective existing methods. However using all frequency sub-bands in decomposition for classification may increase space and time complexity of classification algorithms. In this paper sub-bands with high energy and entropy are only considered for classification to reduce the overheads due to space and time. Fractal dimensions are used to select significant sub-bands for decomposition at each level. Further statistical features of these significant sub-bands are used for classification. This paper describes iris texture classification using selective sub-bands of wavelets based on fractal dimensions and its results are compared with the other classification methods using conventional features.