STRUCTURED SAR IMAGERY FOR OBJECT CHANGE DETECTION AND EXPLANATION

The objective of this project is to develop methodology and prototype algorithms for object-level change detection and change explanation using structural SAR image data. Structural SAR image is obtained by the filtration, detion and transformation of local pixel-level formations of strong and weak electromagnetic (EM) responses into higher-level geometric and morphological micro-elements. These micro-elements form an image of a new type. Change detection is applied to an image of micro-elements. The explanation of a change is deducted from the geometric and morphological comparison. Three algorithms will be developed for (1) the transformation of a pixel-level SAR image in to a geometric SAR image, (2) comparison of object states and clutter characteristics using structural and morphological micro-elemnts, and (3) an explanation of the change in terms of structural similarity/differentiation. The developed methodology will be validated through three groups of experiments. We have already shown that the proposed SAR image transformation is not sensitive to illumination level nor is it sensitive, to a certain degree, to small changes in object pose.