Extracting Features from Point Set Model

This paper present an method for feature extraction from point set. Our algorithm use principle curvatures to flag potential feature points. Using an improved weight sensitive moving least squares, we developed a new approach to detect potential feature curves. The potential feature points are enhanced by projecting the points onto the local potential feature curves. Then smooth the projected points by employing an optimized principal covariance analysis approach. Finally achieve smooth feature curves after resolving gaps and relaxing the results.