Structural and Statistical Feature-Processed PST for Angle Robust Iris Recognition

Abstract

Iris is one such biometric feature that cannot be altered by the individual. In this paper, an improved structural and statistical feature-based iris authentication system is investigated. The model first used the Fisher face approach to generate the structural features. These structural features are processed on block-specific N PST divisions to achieve the directional and geometric variations. The statistical contrast, correlation, energy, and homogeneity features are extracted for each PST block. N-angle-specific M features acquisition method has provided the wider and descriptive feature set for iris images. This statistical transformed dataset is processed by SVM classifier to recognize the iris accurately. The proposed novel model is applied on CASIA-Iris-V3 dataset. The implementation results are applied on angle-variant samplesets. The implementation results identified that the model has achieved the higher accuracy gain for angle robust iris recognition.