Robust Denoising of Point-Sampled Surfaces

摘要

Based on sampling likelihood and feature intensity, in this paper, a feature-preserving denoising algorithm for point-sampled surfaces is proposed. In terms of moving least squares surface, the sampling likelihood for each point on point-sampled surfaces is computed, which measures the probability that a 3D point is located on the sampled surface. Based on the normal tensor voting, the feature intensity of sample point is evaluated. By applying the modified bilateral filtering to each normal, and in combination with sampling likelihood and feature intensity, the filtered point-sampled surfaces are obtained. Experimental results demonstrate that the algorithm is robust, and can denoise the noise efficiently while preserving the surface features.