Construction of biological surface models from cross-sections image processing

An approach improving on existing techniques is presented for blending cross-sections of biological objects to produce a polynomial surface model. As intermediate steps to the final surface skinning, representative data points on the cross-sections are selected for defining piecewise cubic B-splines providing an immediate reduction in storage and computational requirements for the contour representation of the objects. A mesh of quadrilateral patches is subsequently formed over adjacent cross-sections using bicubic B-spline surfaces which exhibit second parametric derivative continuity. The surface model provides a complete and robust representation with significant data reduction. The resulting algorithm is demonstrated using bone data for a human hand.