Low-Dimensional Spectral Feature Fusion Model for Iris Image Validation

Abstract

Iris images are a primal source of biometric security system. Iris images are used as input details in providing access to secure information and operating systems. In the development of biometric security through iris detection, algorithms were developed for extracting multiple features to improve accuracy. Methods were also developed toward validation of iris image in the detection of live or fake samples. Wherein focus tends more toward accuracy, less effort is observed for delay performance and overhead. Feature overhead impacts the system performance. In this paper a feature fusion representing spectral energy and Harlick features is proposed. An approach for dimensional reduction using band spectral correlation with iris image is proposed. The accuracy of the developed approach is validated with the conventional approaches.