The semantic segmentation network performance of large-scale outdoor point clouds is usually limited by the number of input point clouds. In the application of most methods, the point cloud is cut into small pieces as the training input, which will not only lead to heavy preprocessing burdens but also destroy the overall geometric structure of the scene. The transformer network has demonstrated remarkable advantages of attention mechanisms in focusing on crucial features and improving model performance. However, its training and inference on large-scale data are confined by computational complexity. To eliminate these challenges, the attention mechanisms are improved to enhance their performance in processing large-scale input data, while removing limitations imposed by computational complexity. Moreover, a novel local feature enhancement (LFE) module is developed to construct the LFE-Net, which can accurately and efficiently extract spatial and attribute features from local point clouds. In particular, an improved attention module, which is called simple offset attention (SOA), is adopted for local point cloud spatial feature learning. Compared with self-attention, SOA requires less memory and can better capture the fine-grained local features. Furthermore, to effectively avoid the destruction of object geometry and diminish the impact of sample imbalance, a training sample collection method based on the number of different classes is designed. To validate the effectiveness of this method, some experiments are conducted based on publicly accessible outdoor point cloud datasets. The results demonstrate that the LFE-Net can achieve substantial improvements compared with other cutting-edge network models.

The centers of matching craters can be beneficial additions to the control point database. Crater matching on heterogeneous surfaces is helpful in testing its applicability to the entire moon. Therefore, we propose a coarse-to-fine crater matching method for heterogenous surfaces on images acquired from the narrow angle camera (NAC) of the lunar reconnaissance orbiter camera (LROC) and the Chang’e-2 digital orthophoto map (DOM). First, we perform coarse matching based on the Hausdorff distance using the area and coordinates of the crater. Then, the mismatched craters are removed by using the affine transform model fitted by corresponding points of mutual information matching. Finally, we use the retained matched crater centers to fit the affine transformation model between the images, predict the corresponding position, and obtain the corresponding crater around it to achieve fine matching. The results show that the proposed method obtains numerous crater matches on images covering different terrains and solar altitude angles compared to the Hausdorff distance-based crater matching method. For the five experimental scenes registered using matched craters, the mean values of the checkpoints are approximately 2 and 3 pixels for scenes with small and large differences from Chang’e-2 solar altitude angles, respectively, and the standard deviations (STDs) for both are approximately 1 pixel. In addition, the highlands have lower accuracy than the maria, with a variance of less than 1 pixel. Furthermore, the registration accuracy is related to the diameter and number of craters.

The reconstruction of 3-D models of power pylons from light detection and ranging (LiDAR) data plays an important role in power transmission safety. However, accurate reconstruction of power pylon models still faces challenges, e.g., complex structures, missing data, and occlusion. In this letter, a novel four-component segmentation method is proposed for reconstructing power pylon models. In the proposed method, the pylon components of the pylon head, pylon body, cross-arms, and pedestal are first defined in terms of the common features and functionality of each component. Then, these four components are each segmented and identified based on their position and shape features from the raw point cloud. An improved approach based on Metropolis–Hastings sampling and a simulated annealing algorithm is proposed to estimate the model parameters. Based on the estimated parameters, the 3-D shapes of the individual components are reconstructed and stacked to form the whole pylon model. Experimental results show that our methods are able to reconstruct pylons with complex-shaped heads and multiple cross-arms, with an average reconstruction error of less than 0.3 m. Compared with the standard Metropolis–Hastings algorithm with annealing, the parameter estimation process in our strategy improves the computational efficiency by 7.54%.

Rocks are one of the major surface features of Mars. The accurate characterization of the chemical and mineralogical composition of Martian rocks would yield significant evolutionary information about relevant geological processes and exobiological exploration. Many existing rock recognition systems generally assume that all testing classes are known during training. Over real planetary surfaces, the autonomous recognition system is likely to encounter an unknown category of rock that is crucial to the performance of the rock classification task. Therefore, we develop an open-set Martian rock-type classification framework based on their spectral signatures, with the subgoal of new/unknown rock-type recognition and category-incremental learning for expanding the recognition model. First, the spectral signatures of rock samples are captured to characterize their mineralogical compositions and physical properties, which serves as the input of the developed framework. To further produce the highly discriminative feature representation from the original spectral signatures, a Transformer architecture integrated with contrastive learning is constructed and trained in an end-to-end manner to force instances of the same class to remain close-by while pushing those of a dissimilar class farther apart. Following this, according to the extreme value theorem (EVT), category-specific distance distribution analysis is conducted to detect and identify new/unknown types of rock samples due to the isolated characteristics of new/unknown rock samples in the latent feature space. Finally, the recognition model is incrementally updated to learn these identified "unknown" samples without forgetting previously known categories when the associated labels are progressively obtained. The multispectral camera, a duplicated payload of the counterpart onboard the Zhurong rover, is used as the multispectral sensor for capturing the spectral information of the laboratory rock dataset shared by the National Mineral Rock and Fossil Specimens Resource Center for both qualitative and quantitative evaluation. Experimental results indicate the effectiveness and robustness of the developed in situ analysis framework.

Airborne light detection and ranging (LiDAR) point clouds have become growingly popular as a reliable data source for 3-D digital building model reconstruction. Therefore, we develop a label-constraint approach for automatically detecting building roofs using airborne LiDAR point clouds and multispectral images, where the label information is introduced in both the discriminative feature space generation and the detection procedure. To obtain a robust and highly discriminative descriptor, a supervised sparse coding-enhanced bag of visual word (SC-BOVW) model based on a learned discriminative dictionary is used to encode local geometric and spectral information within each super-voxel into high-level semantic representation, which is then fed into a support vector machine (SVM) classifier for distinguishing buildings from others. Additionally, a graph cut-based procedure is used as a postprocessing step to guarantee the spatial consistency in detection results. Experiments were conducted on the International Society for Photogrammetry and Remote Sensing (ISPRS) benchmark data sets. Results indicate that the proposed method is accurate and efficient in terms of building roof region detection. Moreover, the proposed method is superior to other existing methods with average differences in recall of 2.23%, precision of 0.28% and quality of 1.99%.

Global geo-reference Lunar Reconnaissance Orbiter Camera-wide angle camera (LROC-WAC) mosaic imagery provides the precise geographic information for the mapping of mineral elements based on Chang'E-1 interferometric imaging spectrometer (IIM) imagery. However, the traditional image registration methods fail to achieve the accurate registration in between due to heterogeneous characteristics. Therefore, this letter proposes a semiautomatic registration method for Chang'E-1 IIM imagery based on global geo-reference LROC-WAC mosaic imagery. Due to the lack of ground control points, the method implemented a random sample consensus (RANSAC)-guided affine transformation (AT) model to help predict the potential coarse correspondence. Afterward, a multiwindow image matching approach is performed for the fine correspondence. To verify the performance of the proposed method, experiments were performed using Chang'E-1 IIM imagery and geo-reference LROC-WAC mosaic imagery. Experimental results indicate that the proposed method can obtain a massive number of homologous points while minimizing manual intervention, which is comparable to manual results and has high image registration accuracy.

Point cloud feature extraction as a classification task is crucial in maximizing the efficient downstream applicability of raw point clouds. With the goal of learning optimum features for efficient classification of a multi-class point cloud for downstream applications, this letter presents a supervoxelFisher vector (FV)-based approach for airborne light detection and ranging (LiDAR) data classification. In our approach, FV encoding is implemented to deduce compact global descriptors from aggregated supervoxels to establish a more descriptive and discriminative representation, transforming the low-level visual features into high-level semantic features. As a result, the proposed approach combines local and global feature properties through the quantization and aggregation of higher order statistics to harnesses their combined advantages for producing good classification results. Experiments were conducted on the international society for photogrammetry and remote sensing 3-D semantic labeling benchmark data set. Results indicate that the proposed approach is robust and efficient, attained the third best position with an overall accuracy of 81.79%, and ranked first with an F1-score of 72.31%.

The future International Lunar Research Station (ILRS) is to establish autonomous and unattended lunar surface infrastructure with long term energy supply for continuously carrying out lunar scientific research, technical verification and resource development and utilization. Landing site characterization is of great importance to ILRS. For this reason, we conducted a regional geological analysis of the crater fill, including the mapping of stratigraphic units, as well as their age determination with crater-size-frequency distributions (CSFDs) [1].

The pole-like object detection is of significance for robot navigation, autonomous driving, road infrastructure inventory, and detailed 3-D map generation. In this letter, we develop a skeleton-based hierarchical method for automatic detection of pole-like objects from mobile LiDAR point clouds. First, coarse extraction of building facades is adopted for the occlusion analysis. Second, slice-based Euclidean clustering algorithm is implemented to derive a set of pole-like object candidates. Third, skeleton-based principal component analysis shape recognition is presented to robustly locate all possible positions of pole-like objects. Finally, a Voronoi-constrained vertical region growing algorithm is proposed to adaptively producing the individual pole-like objects. Experiments were conducted on the public Paris-Lille-3-D data set. Experimental results demonstrate that the proposed method is robust and efficient for extracting the pole-like objects, with average quality of 90.43%. Furthermore, the proposed method outperforms other existing methods, especially for detecting pole-like objects with a large radius.

Lunar impact craters form the basis for lunar geological stratigraphy, and small-scale craters further enrich the basic statistical data for the estimation of local geological ages. Thus, the extraction of lunar impact craters is an important branch of modern planetary studies. However, few studies have reported on the extraction of small-scale craters. Therefore, this paper proposes a coarse-to-fine resolution method to automatically extract small-scale impact craters from charge-coupled device (CCD) images using histogram of oriented gradient (HOG) features and a support vector machine (SVM) classifier. First, large-scale craters are extracted as samples from the Chang'E-1 images with spatial resolutions of 120 m. The SVM classifier is then employed to establish the criteria for classifying craters and noncraters from the HOG features of the extracted samples. The criteria are then used to extract small-scale craters from higher resolution Chang'E-2 CCD images with spatial resolutions of 1.4, 7, and 50 m. The sample database is updated with the newly extracted small-scale craters for the purpose of the progressive optimization of the extraction. The proposed method is tested on both simulated images and multiple resolutions of real CCD images acquired by the Chang'E orbiters and provides high accuracy results in the extraction of the small-scale impact craters, the smallest of which is 20 m.

Three-dimensional laser scanning can acquire point cloud data with high spatial resolution. However, for practical applications, such as point cloud fitting and 3-D reconstruction, there is usually significant data redundancy, which reduces the operational efficiency. In this letter, we propose a fast point cloud fitting algorithm that uses point cloud simplification to preserve feature boundaries and threshold-independent Bayesian sampling consensus (BaySAC) to fit planar features. We first extract the point features, such as corner points and contour points, using a smoothing analysis of the vicinities of scattered points and an angle analysis of vectors based on search points and their adjacent points. Then, keeping all the feature points, we thin the nonfeature points by constructing a cube grid. Finally, based on the least median squares and the BaySAC algorithm, we propose a robust nonthreshold-dependent method to perform the rapid fitting of planar features in the point cloud after thinning. We used three sets of point cloud data acquired using a 3-D laser scanner to verify the accuracy and efficiency of the planar feature fitting method. The experimental results indicate that the method can extract finer planar features and has significantly better accuracy and computational efficiency than the classical random sample consensus algorithm for the fitting of planar features without using a threshold.

Fitting primitives is of great importance for remote sensing applications, such as 3-D modeling and as-built surveys. This letter presents a method for fitting primitives that fuses the Bayesian sample consensus (BaySAC) algorithm with a statistical testing of candidate model parameters for unorganized 3-D point clouds. Instead of randomly choosing initial data sets, as in the random sample consensus (RANSAC), we implement a conditional sampling method, which is the BaySAC, to always select the minimum number of data required with the highest inlier probabilities. As the primitive parameters calculated by the different inlier sets should be convergent, this letter presents a statistical testing algorithm for the histogram of the candidate model parameter to compute the prior probability of each data point. Moreover, the probability update is implemented using the simplified Bayes formula. The proposed approach is tested with the data sets of planes, tori, and curved surfaces. The results show that the proposed optimized BaySAC can achieve high computational efficiency (five times higher than the efficiency of the RANSAC for fitting a subset of 12 500 points) and high fitting accuracy (on average, 20% higher than the accuracy of the RANSAC). Moreover, the strategy of prior probability determination is proven to be model-free and, thus, highly applicable.

Labeling plays an important role in map production, attaching specific texts to related geographic elements to provide clear environmental references. In three-dimensional geographical information systems (3DGISs), however, cluttering happens fairly commonly because of the unexpected overlapping and occlusion among labels and related objects, and results in an ambiguous and obscure environment. It generally also takes large computing power and memory to visualize spatial entities. Aimed at both unambiguous and efficient 3D map display, this article proposes an adaptive multi-resolution labeling method to deal with point, polyline, and polygon features labeling in a 3D landscape. It implements adaptive placement and view-driven label filtering without obscuring other visual features. The experiments indicated that the display of overlapping labels and label popping are reduced significantly with less computation burden while retaining the rendering quality.