Ellipse fitting for imaged cross sections of a surface of revolution

• We accurately fit the imaged cross sections of a surface of revolution.
• The ellipses can be fitted under the circular point and orthogonality constraints.
• An effective optimization is elaborately designed.
• The method can serve for reconstruction and localization of a surface of revolution.

This paper addresses the problem of accurately fitting the elliptical projections of the cross sections of a surface of revolution (SOR) with the given intrinsic matrix of a camera. By the new approach proposed in this paper, the image points of the SOR cross sections are fitted under the two geometric constraints which are derived from the configuration characteristics of SOR. It is demonstrated by comparison with other previous methods that our method can fit the elliptical projections of SOR cross sections more accurately and robustly. In this paper, we also describe the applications of the algorithm in Euclidean reconstruction of SOR and pose estimation for SOR-shaped object, such as spacecraft.