Comparison of selected features for target detection in synthetic apertureradar imagery

Several methods are available that capture the statistics of radar imagery. The best features, in the sense of man-made target discrimination, are exp ected to be different for different types of natural background and for dif ferent objects of interest such as vehicles. We demonstrate that discrimina tion of natural background and man-made objects using low resolution synthe tic aperture radar imagery is possible using singular value decomposition; several other simple features are also used to augment the feature vector. We use a subset of eigenvectors as features for target discrimination. The optimal set of features used to classify a region as "background clutter on ly" or "target region" is automatically chosen by a standard suboptimal fea ture selection algorithm. (C) 2000 Academic Press.