SAR Image Segmentation based on Multifractal Features

CRISTIAN PACHECO; JULIANA GAMBINI; CLAUDIO DELRIEUX

Bahía Blanca

Workshop; XVIII Reunión de trabajo en Procesamiento de la Información y Control; 2019

Universidad Nacional del Sur - IIIE CONICET UNS

Synthetic Aperture Radar (SAR) imaging is basedon airborne or satellite active microwave sensors that can capturethe earth surface by emitting a signal and receiving the backscat-tered signal that forms the resulting image. Since microwaveradiation is not interfered by sunlight and can pass throughclouds, SAR imagery can be generated oblivious to weather anddaylight conditions. However, the active nature of the imagingprocess determines that SAR images are contaminated by aninherent speckle noise that may degrade signi cantly the qualityand usefulness of the images, and speci c noise-removal processesmay also lter out relevant textural information. In this article,we propose a texture-based method that can be applied for regionsegmentation in SAR imagery. The method is based on localanalysis of the multifractal spectrum and a clustering procedure.The outcomes obtained both with synthetic and real SAR imagesshow better region segmentation results than with state-of-the-artproposals.