Multi-Matching Process Based on Wavelet Transform for Iris Recognition

As a most important biometric solution for personal identification, iris recognition has received increasing attention in recent years. Based on the features of the stochastic iris textural information and local time-frequency properties of the wavelet transform, this paper proposes a new method for iris recognition using a wavelet-based multi- matching system, which includes a coarse matching and its refinement. For the low-frequency components of the wavelet multi-resolution decomposition, a statistical correlation approach is adopted for a coarse matching so that well-matched iris templates in a database can be picked out and need to be further identified. The wavelet modulus maxima are used to locate the sharp variation points, which are taken as iris feature points. A similarity measure based on the distances between the best-matched feature point pairs is taken for the refinement of the coarse matching. Our method well employs the iris morphological characteristics and the multi-matching process to improve the efficiency. The experimental results indicate the validity of the proposed method.