A denoising method for multi-beam point cloud data considering terrain characteristics

Abstract

The invention discloses a method for denoising multi-beam point cloud data in consideration of terrain characteristics. The basic idea of ​​the method is: to establish the topological relationship between point cloud data based on the KD index tree, and to simulate the near-neighborhood data of each point based on the RANSAC algorithm. Fit the local planes, calculate the distance from the point cloud to the respective local fitting planes, and denoise based on the statistical analysis method. In addition, before denoising, the obvious outlier planes are removed according to the normal vector characteristics of the adjacent planes, and the points on the steep slopes are retained. cloud to prevent excessive denoising. Through the above method, the present invention can remove the near-surface noise and obvious outlier noise data in the multi-beam point cloud data, and at the same time better retain information such as edges, optimize the design scheme on the basis of ensuring the above effects, and improve execution efficiency.

Description

technical field [0001] The invention relates to a method for denoising multi-beam point cloud data in consideration of terrain characteristics. Background technique [0002] The shipborne multi-beam bathymetry system can quickly obtain point cloud data containing information such as three-dimensional coordinates and echo intensity on the seabed surface, providing guarantee for generating high-precision DEM. Before building a DEM, the point cloud data must be denoised. Due to the diversity and complexity of the seabed topography, the noise is attached to the surface and is difficult to remove. Manually processing massive data consumes a lot of storage space and computing time, and may even be "erroneously deleted". At present, there are many studies on point cloud data denoising at home and abroad, but there are relatively few studies on multi-beam point cloud data with more noise. The existing technical solutions are difficult to remove near-surface noise, and cannot guaran...

AI Technical Summary

Problems solved by technology

Disadvantages: the calculation of filter factors is complicated, and the threshold is difficult to define

[0005] The third is Gaussian curvature filtering. The principle of this implementation scheme is: calculate the curvature value of the point cloud everywhere, and adopt different filtering schemes according to the change of the point cloud. The effect is better for sharp areas, but the curvature calculation of the point cloud It is more complicated. For the measured multi-beam point cloud data with more noise, the efficiency is low and the expected effect cannot be achieved.

Disadvantages: This type of method is difficult to distinguish near-surface noise data, and has high requirements for the selection of the threshold value. If the selection is looser, it is difficult to remove the noise well. The setting is harsh, which is easy to cause excessive removal, and for sparse non-noise points, it is difficult to distinguish

[0008] 1) Limitation of effect: the existing schemes are difficult to better distinguish the noise data near the surface, or cannot ensure that the terrain data is well preserved while removing the near-surface noise data; 2) Poor adaptability to terrain: the existing schemes are respectively applicable to Different point cloud data, difficult to adapt to terrain changes

3) It is difficult to achieve a balance between the denoising effect and the efficiency of the method: the method with high efficiency is difficult to guarantee the effect, and the method with better effect is often complex in calculation and difficult to guarantee the efficiency

Images