Iris Recognition Based on Improved Compressive Sensing Algorithm
Yongjun Li · Journal of Information and Computational Science · 2014
The traditional iris recognition algorithm exists some shortcomings, such as the higher iris feature dimension, and which caused low efficiency of algorithm execution. A quick iris recognition method based on improved Compressive Sensing (CS) algorithm is proposed in this paper. The conventional construction process of redundant atom library is just based on image gray information. It converts two-dimensional image into a one-dimensional column vector, ignoring the image texture informations. There is a high dimension of the shortcomings of one-dimensional vector. In this paper, a redundant atom library construction algorithm is proposed. It is based on Gabor + PCA (GPCA) to retain the image texture features, greatly reduces the dimension of the image at the same time. Orthogonal Matching Pursuit (OMP) algorithm is used to achieve the sparse coefficient, and then identify the image. The experimental results show that: the iris recognition algorithms based on GPCA and CS can reduce the dimension of the image features greatly, and execution speed and recognition accuracy is high.