Iris recognition algorithm based on feature weighted fusion

The features of single iris are relatively simple, which can easily lead to the inaccurate iris recognition. To solve this problem, the feature weighted fusion is used to represent the iris texture in this paper. First, the iris texture spatial domain features and frequency domain features are extracted using Principal Component Analysis (PCA) to reduce noise and redundancy. Second, Dichotomous Statistical Local Binary Pattern (DSLB) is used to represent variation rule of iris texture gray value, forming spatial domain feature code; and for frequency domain features, 2D-Haar wavelet is used to extract the high frequency coefficients of iris feature and form frequency domain feature code. Third, the Hamming distances of the two feature codes are calculated respectively and multiplied by the corresponding weighted factors. Finally, the iris category is determined by comparison with the set classification threshold. A variety of iris libraries are used to compare the performance of proposed algorithm with other iris recognition algorithms. Experiment results show that the proposed algorithm has more advantages in recognition rate, equal error rate and stability. © 2019, Editorial Board of Jilin University. All right reserved.