Point Set Denoising Using Bootstrap-Based Radial Basis Function.

Area of Science:

• Computer Vision
• Geometric Modeling
• Numerical Analysis

Background:

• Noisy data is a significant challenge in 3D scanning and point set modeling.
• Accurate surface smoothing and denoising are crucial for preserving geometric features.
• Radial basis functions (RBFs) are effective for surface approximation but require careful parameter selection.

Purpose of the Study:

• To develop and evaluate a novel smoothing algorithm for point set denoising using RBFs.
• To improve the accuracy of surface smoothing by incorporating bootstrap test error estimation.
• To preserve fine geometric details during the denoising process.

Main Methods:

• Application of bootstrap test error estimation for selecting smoothing parameters of RBFs, specifically thin-plate splines.
• Development of a smoothing algorithm utilizing a bootstrap-based RBF.
• Integration of a k-nearest neighbour search for point set projection onto the approximated surface.

Main Results:

• The proposed bootstrap-based RBF method effectively denoises point sets.
• Geometric features of the surfaces are well-preserved after applying the smoothing algorithm.
• The method demonstrates competitive or superior performance compared to existing smoothing techniques.

Conclusions:

• Bootstrap test error estimation is a viable approach for optimizing RBF smoothing parameters.
• The proposed algorithm offers an effective solution for point set denoising in 3D scanning applications.
• This method successfully balances noise reduction with feature preservation.