Residual-Sparse Fuzzy <i>C</i>-Means Clustering Incorporating Morphological Reconstruction and Wavelet Frame

In this article, we develop a residual-sparse Fuzzy C -Means (FCM) algorithm for image segmentation, which furthers FCM's robustness by realizing the favorable estimation of residual (e.g., unknown noise) between an observed and its ideal version (noise-free image). To achieve sound tradeoff detail preservation noise suppression, morphological reconstruction is used to filter image. By combining filtered images, weighted sum generated. Tight wavelet frame decomposition transform into corresponding feature set. Taking such set as data clustering, impose $\ell _0$ regularization term on objective function, thus resulting in FCM, where spatial information introduced improving making more reliable. further enhance segmentation accuracy proposed employ smoothen labels generated clustering. Finally, based prototypes smoothed labels, segmented reconstructed using tight reconstruction. Experimental results regarding synthetic, medical, real-world images show that effective efficient, outperforms peers.