Method of combining two sets of geometric data for a given shape are proposed. The data can come from the structured light system and/or from the boundary extraction system based on "Silhouette". The basic approach is to match the geometries represented by the two data sets and then to use a weighted combination of two corresponding points. Since the geometry from the structured light system has a better accuracy in the positional data, while the other system in the smoothness of tangential vectors, the algorithm have been designed to fully utilize these characteristics.
To start the algorithm, the two data sets are first rearranged to have the same data form containing both positional and tangent informations. To identify the orientation of represented geometries and to match them, informations on the first and the second moment of inertia are used. To find the points in correspondence, the intersection points of on geometry by some imaginary light rays emanating from the rotation axis and passing through the data points of the other data set are found either by using triangular patches or by curved surface patches. An iterative method based on a Taylor series expansion in u and v has been proposed for the latter case.
As an efficient tool of geometric data manipulation, a refined data by zooming operation can be used to replace or fill in a subset of a given shape. The software system developed has been tested extensively with different shapes such as pentagon, bottle, foot, TV bulb, Agrippa and so on. The accuracy is estimated to be within 0.5mm for these cases.
