Removal of LiDAR Negative Outliers Based on Retroreflective Surface
LiDAR is an essential tool for terrain data acquisition; however, its application in coastal environments is often limited by negative outliers caused by multipath reflections. The negative outliers can result in deviations of several meters, significantly complicating subsequent data processing and...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11066278/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | LiDAR is an essential tool for terrain data acquisition; however, its application in coastal environments is often limited by negative outliers caused by multipath reflections. The negative outliers can result in deviations of several meters, significantly complicating subsequent data processing and analysis. This article investigates the retroreflective characteristics of negative outliers in terms of spatial structure and intensity and presents a negative outlier removal algorithm based on these features. First, the LiDAR surveying equation is introduced to establish the intensity relationship between negative outliers and their corresponding preliminary reflection points. Second, by analyzing the spatial distribution of point clouds, a covariance matrix is generated, and eigenvalue decomposition is performed to extract structural descriptors for identifying outliers. Third, a terrain mesh model is constructed to approximate the retroreflective surface, enabling a feature-based comparison between negative outliers and their preliminary reflection points. Finally, points below the terrain mesh and their corresponding reflection points are extracted. By comparing their structural similarity and intensity relationships, negative outliers are accurately identified and removed. Experimental results validate the effectiveness of the proposed algorithm, achieving a precision of 88.97% and a recall of 91.94%, ensuring robust outlier removal while preserving terrain details. |
|---|---|
| ISSN: | 1939-1404 2151-1535 |