GEOMETRIC FEATURE DETECTION FOR REVERSE ENGINEERING USING RANGE IMAGING

The use of range imaging has been gaining popularity in reverse engine ering. One challenging task is the detection of feature information fr om range images. In this paper, an approach to detect geometric featur es from range images using a fuzzy partitioning theory and geometric i nvariants is developed. Based on the fuzzy C-shell clustering techniqu e, quadric features are partitioned into primitive clusters. Instead o f performing sequential model fittings, general quadric surfaces as ob ject shells are fitted concurrently, The geometric representations of prototypes are generated during the above process of pattern classific ations. The integration of the partition with the invariant analysis m akes it possible to detect geometric features from depth maps for the development of reverse engineering. (C) 1996 Academic Press, Inc.