Search

Point cloud registration is crucial for various applications such as robotics and urban planning, yet it remains challenging due to issues such as noise, variable resolutions, and uncertainties in the initial pose. The normal distribution transform (NDT) algorithm has become a standard approach for registering point scans, but its application to large-scale scenes faces significant challenges due to inherent limitations. These include computational complexity, susceptibility to noise interference, limited adaptability to varying voxel resolutions, and substantial initial pose deviations. To address these issues, we propose the multiscale iterative NDT (MI-NDT) through a meticulously designed multiscale iterative optimization framework. The MI-NDT algorithm voxelizes the reference scan into multiscale voxels based on the optimal flatness ratio criterion. At each scale, the classic NDT algorithm computes the pose of the source scan. The pose obtained from coarse voxels serves as input for the next optimization step on finer voxels. This iterative refinement continues until the finest voxel level, where the resulting pose corresponds to the optimal translation parameters. The experimental results demonstrate the effectiveness of the MI-NDT algorithm in terms of registration accuracy and robustness. Comparative experiments with existing algorithms show significant improvements in accuracy, especially under varying voxel resolutions and noise levels. Moreover, the algorithm exhibits enhanced tolerance to initial pose deviations, enabling successful registration even with poor initial poses.

The global bathymetry models are usually of low accuracy over the coastline of polar areas due to the harsh climatic environment and the complex topography. Satellite altimetric gravity data can be a supplement and plays a key role in bathymetry modeling over these regions. The Synthetic Aperture Radar (SAR) altimeters in the missions like CryoSat-2 and Sentinel-3A/3B can relieve waveform contamination that existed in conventional altimeters and provide data with improved accuracy and spatial resolution. In this study, we investigate the potential application of SAR altimetric gravity data in enhancing coastal bathymetry, where the effects on local bathymetry modeling introduced from SAR altimetry data are quantified and evaluated. Furthermore, we study the effects on bathymetry modeling by using different scale factor calculation approaches, where a partition-wise scheme is implemented. The numerical experiment over the South Sandwich Islands near Antarctica suggests that using SAR-based altimetric gravity data improves local coastal bathymetry modeling, compared with the model calculated without SAR altimetry data by a magnitude of 3.55m within 10 km of offshore areas. Moreover, by using the partition-wise scheme for scale factor calculation, the quality of the coastal bathymetry model is improved by 7.34 m compared with the result derived from the traditional method. These results indicate the superiority of using SAR altimetry data in coastal bathymetry inversion.

Extracting pylons from point clouds gathered by unmanned aerial vehicle LiDAR systems (UAVLS) is challenging, particularly in complex environments. The difficulty arises due to the presence of nearby objects and vegetation, which are intertwined around the pylon legs and structures. In such situations, it is hard to identify pylon points from vegetation points using spatial point-level features. To overcome this issue, we propose a classification approach using the random forest (RF) classifier incorporating smoothed color features. The procedure involves the following steps: first, a set of local geometry and distribution features are generated by combining spherical and cylindrical neighborhoods. The optimal neighborhood radius is then determined by experimenting with radii ranging from 1 to 8 m. To further enhance the method’s accuracy, we introduce smoothed color features within a neighborhood to eliminate the misclassification points of the vegetation close to the pylon leg. The results indicate that smoothed color features outperform intensity and single-color features, achieving a high extraction rate of 98.65% and 99.99% for the two datasets, respectively. Furthermore, our method is robust against varying levels of noise and density, and it significantly improves pylon classification accuracy when coupled with other classifiers such as support vector machine (SVM) and AdaBoost.