Low-complexity iris recognition with oriented wavelets

Abstract

Iris recognition recently became an active field of biometric security because of reliability and easy non-invasive acquisition of the data. The randomness and stability of the iris textures allow for a convenient application in personal authentication and identification. In a novel iris recognition method presented here, the iris features are extracted using the oriented separable wavelet transforms (directionlets) and they are compared in terms of a weighted Hamming distance. The feature extraction and comparison are shift, size and rotation-invariant to the location of iris in the acquired image. The generated iris code is binary, whose length is fixed (and therefore commensurable), independent of the iris image, and comparatively short. The novel method shows a good performance when applied to a large database of irises and provides reliable identification and verification. At the same time, it preserves conceptual and computational simplicity and allows for a quick analysis and comparison of iris samples.