Calculation of the Normal Vector using the 3 × 3 Moving Mask Method for Freeform Surface Measurement and its Application

Abstract

The concept of a 3 3 3 moving mask operation image-processing technique is proposed to calculate the normal vector of measuring points in this study. The method developed reduces greatly the calculation time of matrix operation and memory space in comparison with the traditional composite Ferguson-spline method. The methodology for calculating the normal vector is to select eight neighbouring points at equal distances in the vicinity of an arbitrary node on the surface, from which a small surface patch can then be constructed from the nine selected points. Different analytical methods are used to calculate the unit normal vector, namely the Bezier method with uniform parameters and the Bezier method with non-uniform parameters, and are discussed in this study. The accuracy of these two methods in calculating the unit normal vector was also verified by calculating different positions on a spherical surface. The Shepard interpolation method was adopted to interpolate a few control points from a massive number of measured data points to establish the CAD model of a freeform surface using a rectangular grid. The method developed was applied for the measurement of a freeform surface (mouse surface) using a coordinate measuring machine. The local Shepard interpolation method was used to interpolate 16 control points from 1054 measured data points. A bi-cubic Bezier- and B-spline surface CAD model were constructed through these interpolated control pointer.