Wavelet-Based Noise Removal Technique for Remotely-Sensed Data

Abstract

Filtering of the LiDAR data is challenging due to the complex distribution of the surface and the various types of contaminating noises. Also the collected data contain much information that requires the appropriate pre-processing in order to generate good. In this paper a new approach has been proposed for denoising and processing LiDar data. The proposed method utilize the advantages of multiresolution analysis and robust fitting. It has been shown that it excellently removes both additive noise and artifacts with retaining the important parts of the surface model. The method requires only low resolution levels and is able to avoid data loss.