Comparative Study of Automatic Plane Fitting Registration for MLS Sparse Point Clouds With Different Plane Segmentation Methods

Abstract

ABSTRACT:The least square plane fitting adjustment method has been widely used for registration of the mobile laser scanning (MLS) point clouds. The inputs for this process are the plane parameters and points of the corresponding planar features. These inputs can be manually and/or automatically extracted from the MLS point clouds. A number of papers have been proposed to automatically extract planar features. They use different criteria to extract planar features and their outputs are slightly different. This will lead to differences in plane parameters values and points of the corresponding features. This research studies and compares the results of the least square plane fitting adjustment process with different inputs obtained by using different segmentation methods (e.g. RANSAC, RDPCA, Cabo, RGPL) and the results from the point to plane approach -an ICP variant. The questions for this research are: (1) which is the more suitable method for registration of MLS sparse point clouds and (2) which is the best segmentation method to obtain the inputs for the plane based MLS point clouds registration? Experiments were conducted with two real MLS point clouds captured by the MDL -Dynascan S250 system. The results show that ICP is less accurate than the least square plane fitting adjustment. It also shows that the accuracy of the plane based registration process is highly correlated with the mean errors of the extracted planar features and the plane parameters. The conclusion is that the RGPL method seems to be the best methods for planar surfaces extraction in MLS sparse point clouds for the registration process.

Review of mobile laser scanning target‐free registration methods for urban areas using improved error metrics

“…In this case, point‐to‐model matching is called least squares plane‐fitting adjustment (LSPFA), whose goal is to find the six transformation parameters that minimise the sum of the squares of the distances from points to the planar surfaces. According to Nguyen et al. (2017), at least a triplet of planes must exist in the scanning area in order to ensure the success of LSPFA.…”

“…The check planes were also used as the input for the LSPFA process as they are sufficient to meet the requirements for the success of LSPFA (Nguyen et al., 2017). None of the ground planes were used as input as they exhibit slight curving trends, which are considered as unreliable for registration.…”

Landslide Monitoring and Assessment for Highway Retainingwall: The Case Study of Taşkent(turkey) Landslide

“…In theory, the objective of the least square adjustment method is the sum of the squares of the errors times their respective weights minimization for each normal equations (Long Nguyen et al, 2017). Hence, mean error values of each point of planes are computed to explain the differences of the transformation parameters.…”

Reproducibility Experimentation among Computer-Aided Inspection Software from a Single Point Cloud

“…e presented method has a linear algorithmic complexity and it is able to detect large and small planes in very large data sets. In 2017, Nguyen et al [24] presented a comparative study of the least square plane fitting algorithms with different segmentation methods (e.g., RANSAC, RGPL, Cabo, and RDPCA). ey validated the study by two real point clouds collected by a Dynascan S250 scan system.…”