Noise-reduction algorithms for digital radar images and classification of remote sensed agricultural SAR images

Digital techniques for image processing and automatic classification in remote-sensing applications are described. Some digital filters suitable for radar noise reduction are reviewed. A novel filtering algorithm for multiplicative noise rejection is then proposed. Some algorithms using the maximum-likelihood decision rule or the Euclidean distance are applied to preprocessed images for an automatic discrimination of agricultural crops. Microwave images obtained by means of synthetic aperture radars and collected during the European SAR 580 campaign are used to evaluate the performance of the filtering algorithms. Both single-band and two-band classification with a correct classification probability of over 95% were obtained.