Point Cloud Denoising based on Adaptive Wavelet Transformation

The point cloud data obtained by 3D laser scanner are not only simple in structure, easy to operate, but also don't need store topological relationships between points. They can be used to express complex geometry and surface characteristics of irregular objects. However, in the process of obtaining the data, because of many factors, such as the human factor, the change of the environment or the defects of the equipment itself, the data obtained are contaminated by noise. Therefore, point cloud data denoising is an important post-processing step performed on potentially noisy data obtained from a 3D scanner. A new point cloud denoising method is proposed based on adaptive wavelet transformation, which includes three steps: namely, point cloud data decomposition, wavelet coefficients neighborhood adaptive division and wavelet coefficients inverse transformation. The method can not only effectively remove the noise, but also preserve sharp features and surface details. At last, the performance of the proposed method was illustrated with a validation experiment.