OBJECT-ORIENTED FEATURE-TRACKING ALGORITHMS FOR SAR IMAGES OF THE MARGINAL ICE-ZONE

Previous work in tracking sea ice with sequential synthetic aperture radar (SAR) images has described specific algorithms. None of these algorithms is capable by itself of tracking sea ice under all circumstances, given that sea ice naturally exhibits a, wide range of movement and that conditions found in the arctic—i.e., the marginal ice zone—are extremely variable. Instead, these algorithms are capable of tracking sea ice under a limited set of conditions; several must be combined to cover the entire gamut of movement and conditions found in the arctic. The problem comes when a computer must choose and apply the appropriate algorithm: how does it make its choice and know where that choice is most applicable? An unsupervised method is reported that chooses and applies the most appropriate tracking algorithm from among different sea-ice tracking algorithms. In contrast to current unsupervised methods, this newly devised method chooses and applies an algorithm by partially examining a sequential image pair to draw inferences about what was examined. Based on these inferences, the reported method subsequently chooses which algorithm to apply to specific areas of the image pair where that algorithm should work best. © 1990 IEEE