The paper proposes a method of constructing the three-dimensional CAD model from digitized cloud-ofpoints data of the human body, thereby automatically extracting several featured dimensions. The main steps are; segmentation, sectioning, fairing and surface approximation. In segmentation process the input three-dimensional set of points are grouped into each functional part of the human body by searching the boundary points when being viewed from a certain direction. The next step slices the grouped points to obtain section curves which will be faired (smoothed) by the difference-fairing method. Each faired section curve (a set of points) is approximated to the NURBS form by use of well known method, from which we can construct a NURBS surface. The parametric surfaces for functional parts of a human body are interrogated to extract several featured dimensions of a human body.
