Surface Curvature Estimation from Range images Based on Adaptive Surface Fitting and Robust M-Estimation

摘要

Curvature estimation plays an important role in range image analysis. Normally people use finite difference or fit local surface to calculate partial derivatives. In order to get a reliable estimation, we propose an estimation algorithm based on adaptive surface fitting and robust M-estimation. First, we assume the surface of rang-image is not global but piecewise smooth, and estimate curvature from the smooth piece of surface to which the current point belongs. Second, we define an energy function to measure the smoothness of fitting window, then move the center of fitting window adaptively to minimum the energy function. Third, we make a robust M-estimation on the data set to eliminate the effects of the remaining "outliers". Theoretic analysis and Experimental results prove that the estimation algorithm is robust, reliable, efficient and with less computational complexity.