Three Dimensional Boundary Detection Using Higher-Order Surface Fitting and Directional Smoothing

The authors propose an algorithm for detection of three-dimensional bundaries in noisy images based on higher-order polynomial surface fitting and directional smoothing. Fitting a polynomial to the local intensities gives the intensity hypersurface. An isointensity surface i s derived from the hyperplane and directional smwthiag is defined as smoothing along this isointensjty surface. The developed hundary detection algorithm exploits this smoothing technique and gives g d edge preservation, local~zrltion and noise reduction. Experiments were performed on synthetic noisy data sets. Results indicate that the algorithm is superior to orher edge-preserving smoothing approaches reported previously.