Iris Recognition Based on Adjustable Scale Wavelet Transform

Abstract

This paper represents a new method for iris recognition using the bandpass characteristic of wavelets and wavelet transform principles for detecting singularities. The iris image texture is divided into eight bands, and each band is processed to generate a 1-D intensity signal. The appropriate scale wavelet is used to filter each signal for quantization to 256 byte iris feature codes. Then the Hamming distance is adopted to match two iris codes. In particular, the appropriate scale for each signal is searched for by minimizing the Hamming distance. The results show that the performance of this method is better than some algorithms. Meanwhile, it is translation, rotation, and scale invariant, so it is promising to be used in a personal identification system.