APPROXIMATE ORTHOGONAL DISTANCE REGRESSION METHOD FOR FITTING QUADRICSURFACES TO RANGE DATA

Fitting surfaces to 3-D data is one of the basic methods of surface de scription for 3-D vision. Most techniques of surface fitting proposed in the literature are ''least-squares''-based that rarely produce sati sfactory results if noise level is very high or if the data points are sampled from a small area. A new approach is presented in this paper that minimizes the mean squared approximate orthogonal distances with linearization using Newton's iteration method. This approach usually y ields a good fit and the algorithm is reliable and efficient for real applications. Results are reported for synthetic data and several exam ples of real range data. Experimental results demonstrate that the app roximate orthogonal distance performs better than the least squares ba sed methods.