Euclidean Calculation of Feature Points of a Rotating Satellite: A Daisy-Chaining Approach
Keith Dupree, Nicholas Gans, William T. MacKunis, Warren E. Dixon · Journal of Guidance Control and Dynamics · 2008
The occlusion of feature points and/or feature points leaving the field of view of a camera is a significant practical problem that can lead to degraded performance or instability of visual servo control and vision-based estimation algorithms. By assuming that one knownEuclidean distance between two feature points in an initial view is available, homography relationships and image geometry are used in this paper to determine the Euclidean coordinates of feature points in the field of view. A new daisy-chainingmethod is then used to relate the position and orientation of a plane defined by the feature points to other feature-point planes that are rigidly connected. Through these relationships, the Euclidean coordinates of the original feature points can be tracked even as they leave the field of view. This objective is motivated by the desire to track the Euclidean coordinates of feature points on one face of a satellite as it continually rotates and feature points become self-occluded. A numerical simulation is included to demonstrate that the Euclidean coordinates can be tracked even when they leave the field of view. However, the results indicate the need for amethod to reconcile any accumulated errorwhen the feature points return to thefield of view. Nomenclature A = intrinsic camera-calibration matrix dj = distance to j plane along nj F j, Fj = frames attached to the j and j planes Gj = projective homography matrix of the jth frame Hj = Euclidean homography matrix of the jth frame I = fixed coordinate frame attached to the camera mji, m ji = normalized Euclidean coordinate of the ith feature point of the j and j planes expressed in I