The segmentation of synthetic aperture radar (SAR) images is a longstandingyet challenging task, not only because of the presence of speckle, but also dueto the variations of surface backscattering properties in the images.Tremendous investigations have been made to eliminate the speckle effects forthe segmentation of SAR images, while few work devotes to dealing with thevariations of backscattering coefficients in the images. In order to overcomeboth the two difficulties, this paper presents a novel SAR image segmentationmethod by exploiting a multi-scale active contour model based on the non-localprocessing principle. More precisely, we first formulize the SAR segmentationproblem with an active contour model by integrating the non-local interactionsbetween pairs of patches inside and outside the segmented regions. Secondly, amulti-scale strategy is proposed to speed up the non-local active contoursegmentation procedure and to avoid falling into local minimum for achievingmore accurate segmentation results. Experimental results on simulated and realSAR images demonstrate the efficiency and feasibility of the proposed method:it can not only achieve precise segmentations for images with heavy specklesand non-local intensity variations, but also can be used for SAR images fromdifferent types of sensors.
展开▼