Residual-driven Fuzzy C-Means Clustering for Image Segmentation

Due to its inferior characteristics, an observed (noisy) image's direct use gives rise poor segmentation results. Intuitively, using noise-free image can favorably impact segmentation. Hence, the accurate estimation of residual between and images is important task. To do so, we elaborate on residual-driven Fuzzy C-Means (FCM) for segmentation, which first approach that realizes leads participate in clustering. We propose a FCM framework by integrating into residual-related fidelity term derived from distribution different types noise. Built this framework, present weighted $\ell_{2}$-norm weighting mixed noise distribution, thus resulting universal algorithm presence or unknown Besides, with constraint spatial information, becomes more reliable than only considering itself. Supporting experiments synthetic, medical, real-world are conducted. The results demonstrate superior effectiveness efficiency proposed over existing FCM-related algorithms.