Abstract: Matching high accuracy surface process in Class-A surface reconstruction is complicated.It needs a large number of operations,and cannot modify due to do not retain parameters after the surface fitting.There exist many problems which causing surface reverse reconstruction work repeatedly,therefore the thesis put forward a kind of accurate and efficient method for surface reconstruction.According to the discrete point cloud based on K-neighborhood search,using the multi-mode subsection interpolation fitted curvature continuous curves such as line,arc and other ways,retain parameter characteristics to modify conveniently,and then using the NURBS theory to fit surface,ensure the surface fitting accuracy and curvature continuous transition though adjusting the control points and the node vector or other factors.Through the analysis of case,the fitting method of Class-A surfaces is quick and accurate,it retains the shape parameter and improve the accuracy and efficiency of surface reconstruction.

Key words: reverse engineering, class-A surface, G²continuous, surface reconstruction, smoothing