A Feature Preserving Denoising Approach for Scanned Models Based on Trilateral Filtering

With the wide application of 3D scanning technology,how to model the target object based on the discrete scanning data in a robust and efficient way has become one of the research topics in Computer Graphics.This paper proposes a robust filter function,called trilateral filter function,to filter the noises in the domains of surface curvature,normal and position at each sample point while the local surface is fitted in a quadratic form.By optimizing the filtering result of curvature tensor,our approach is capable of preserving the feature on the target model.Since it requests no connection information between the adjacent sample points,the proposed approach is applicable to both mesh and point models.