A Research on MBES data classification denoising algorithm based on Octree index

Abstract

For the Multi-Beam Echo Sounder (MBES) data denoising problem, the noise is divided into near-ground noise and outlier noise away from the seabed. The algorithm in this paper considers the spatial distribution characteristics of noise and removes two kinds of noise by initial denoising and precise denoising. The algorithm first uses the octree index to organize the point cloud and removes the outlier noise. Then combined with Bayesian estimation theory and statistical methods to remove near-ground noise. This paper designs experiments to process MBES data and compare it with the statistical filter of PCL point cloud library. The experimental results show that compared with the PCL statistical filter, the algorithm used in this paper has good denoising effect and can better preserve the information such as regional boundaries.