An algorithm for surface reconstruction from planar contours using smoothing splines

This paper presents a fast algorithm for constructing a smooth three-dimensional surface over a set of cross-sectional contours. We assume that these sections are perpendicular to the z-axis and first consider the case that the surface can be represented in cylindrical coordinates. An approximation is then determined for r(θ, z) by using tensor product splines which satisfy certain boundary constraints. The algorithm is an extension of an existing semi-automatic surface fitting algorithm. The knots of the spline are chosen automatically but a single parameter is expected to control the tradeoff between closeness of fit and smoothness of fit.Both open and closed surfaces can be represented. In particular we demonstrate the use of a non-linear transformation for obtaining smooth closed surfaces.The algorithm can easily be extended to the reconstruction of surfaces which cannot be represented in cylindrical coordinates. A number of applications are also briefly discussed such as the calculation of volumes and the intersection with other surfaces. We have applied the method in practice to obtain a 3-D integrated image of the cerebral blood vessels and CT view of tumor for stereotactic neurosurgery.