A two-dimensional segmentation algorithm for SAR images
L. M. Delves, R T McQuillan, R. Wilkinson, J B E Sandys-Renton, Christopher J. Oliver · Inverse Problems · 1991
The despeckling and segmentation algorithm described by the authors, is extended to two-dimensional images. The method is based upon a least-squares fit of the data to a 'cartoon' model designed to retain the necessary segmentation information within a noise-free framework; the authors describe both the construction of the cartoon, and efficient techniques for implementing the least-squares fitting procedure. Results suggest that this approach has considerable promise for segmenting complex images.