Robust credibilistic fuzzy local information clustering with spatial information constraints

• DOI: 10.1016/j.dsp.2019.102615
• Publication Date: 2019-11-12T11:44:40Z
• Authors: Chengmao Wu; Xiaoqiang Yang
• Keywords: FLAME clustering; Robustness

Abstract Credibilistic fuzzy clustering is a novel data analysis method. Aiming at the shortcoming that credibilistic fuzzy clustering algorithm (CFCM) lacks the ability of noise suppression for image segmentation, a robust credibilistic fuzzy C-means clustering with weighted local information (CWFLICM) is proposed. At first, the fuzzy local information is introduced to guarantee the noise insensitiveness and image detail information. Secondly, the similarity measure constraints of distance between the current pixel and clustering centers is constructed in combination with the membership information, the spatial information, and the intensity information of its neighbor pixels. In addition, an adaptive local information factor in the similarity measure constraints is constructed by using the unhomogeneous degree of neighborhood intensity information. In the end, a kernelized CWFLICM algorithm is also proposed to improve its anti-noise robustness for image segmentation. Many experimental results show that the proposed algorithm has better segmentation performance and stronger anti-noise robustness than existing fuzzy clustering algorithms.