Iris recognition method based on improved sparrow search algorithm and BPNN

Abstract Iris recognition is a very popular and efficient biometric method that has a wide range of applications in finance, security and many other areas. The uniqueness and invariance of the iris makes it highly accurate in identification. Although there are many approaches to iris recognition, such as deep learning based methods, drawbacks such as high computational complexity, high memory requirements, and long training times still exist. Since BP neural networks have the advantages of simple structure and short training time, this paper, an iris recognition method based on improved sparrow search algorithm (CRSSA) optimized back propagation neural network (BPNN) is proposed. The sparrow search algorithm is improved using improved Kent chaos mapping strategy, Dynamic adaptive weight strategy and Logarithmic spiral-based random walk strategy. Then the improved algorithm is used to optimize the BP neural network. Finally, the improved network is used as a classifier for iris recognition. Preprocessing of iris images using Hough transform based processing. The iris image is processed using feature extraction methods based on first order statistical measures and second order statistical measures (F-S) and finally iris recognition is performed using the method proposed in this paper. The experimental results show that the method proposed in this paper exhibits higher recognition accuracy and stability than other algorithms in the experiment on CSAIA-iris-V4 and JLU-4.0 datasets. The improved sparrow search algorithm (CRSSA) proposed in this paper also performs better than the other algorithms in the experiments in terms of convergence accuracy and stability.