Comparing Score Level and Feature Level Fusion in Multimodal Biometric Identification using Iris and Palmprint Traits with Fractional Transformed Energy Content

In previous paper score level fusion performance is tested with various matching score proportions with Iris and Palmprint images. Here feature level fusion is done using fractional energy of transform with Iris and Palmprint biometric traits. The benefit of energy distribution of transforms in higher coefficients is taken here to reduce the feature vector size of image. Iris and Palmprint are together taken here for identification with fractional energy of Slant, Walsh, Haar, Hartley and Kekre, Sine, Cosine transformed Palm and Iris images. The test beds of 60 pairs of Iris and Palmprint samples of 10 persons (6 per person of iris as well as Palmprint and fused Iris: Palmprint) are used as test bed for experimentation. Experimental result in feature level fusion of Iris: Palmprint using Hartley transform with (1:3) threshold gives highest GAR value.