Matching Radar and Satellite Images Employing the Hausdorff Distance for Ship Positioning and Trajectory Estimation

The paper describes a technique to match satellite and radar images using the Hausdorff Distance (HD). Minimization of the average of a truncated array of sorted Hausdorff distances is used to get estimates for the radar location, together with sensor bias errors and its platform speed vector. The technique is applied to maritime navigation, where geo-referenced satellite images (UTM) are matched to radar scan images. Good estimates are obtained for these variables, thus providing an alternative method to obtain a vehicle's location in GPS-denied environments or for indoor autonomous navigation.