Abstract: Aiming at the problem that the traditional fuzzy clustering algorithm ignores the uncertainty of spectral measure in the image segmentation process and the pixel category is non-membership in the clustering process, a remote sensing image segmentation algorithm based on intuitionistic fuzzy set is proposed.Firstly, an intuitionistic fuzzy generator is designed, and the spectral measure uncertainty of images is analyzed by the maximum entropy method, and the spectral measure uncertainty of images is modeled by solving the band index and transforming the remote sensing images into intuitionistic fuzzy sets.Secondly, in the process of clustering, the pixel category membership degree and pixel category non-membership degree are considered simultaneously, and the objective function is defined by combining the distance between intuitionistic fuzzy sets, so as to improve the algorithm's processing ability of category fuzzy information and achieve accurate segmentation of remote sensing images.Finally, the proposed algorithm and the comparison algorithms are used to segment the simulated image and the real color remote sensing image respectively. The qualitative and quantitative evaluation of the segmentation results show that the proposed algorithm can better deal with the uncertainty of the image itself and the clustering process, and obtain higher precision image segmentation results.

Key words: image segmentation, intuitionistic fuzzy sets, non-membership degree, hesitation degree, intuitionistic fuzzy FCM