The purpose of this work was to present a surface fitting algorithm for sweep surface reconstruction from three-dimensional measured data. The sweeping rule considered in this work was essentially translational sweeping in which the generators traverse about the directors to form the desired sweep surface. The sweep-surface fitting was formulated as a nonlinear least-squares minimization problem for which an error expression was minimized which yields the optimized generators, the directors and the parameter values corresponding to each measured data. An algorithm was presented also to convert the sweep surface model into a composite spline surface which could be imported into most CAD/CAM systems. Effective experimental results were provided to illustrate the feasibility of the proposed strategy. (C) 1998 Elsevier Science Ltd. All rights reserved.
