Information granulation-based fuzzy RBFNN for image fusion based on chaotic brain storm optimization

Abstract

Image fusion based on regional feature is a challenging task, which has difficulty in obtaining optimal weight of every image source. In this paper, Information Granulation-based Fuzzy Radial Basis Function Neural Networks (IG-FRBFNN) is utilized to obtain weight of each source image dynamically. In the proposed network, the fuzzy C-means (FCM) clustering is exploited to form the premise part of the rules. Additionally, weighted least square (WLS) learning is adopted to estimate the coefficients of polynomials, which have four types to form the consequent part of the model. Since the performance of IG-FRBFNN is directly affected by some key parameters of the networks, inspired by the chaos theory, chaotic brain storm optimization (CBSO) is proposed in this paper, carrying out the structural and parametric optimization of the network respectively. A series of experimental results demonstrate that the proposed approach performs better compared with the other state-of-the-art approaches.