Point cloud completion is crucial for downstream tasks in 3D visual perception. However, existing methods often struggle to generalize to real-world scans due to their heavy reliance on abundant paired point clouds for training and their neglect of the distribution shift between training and testing datasets. To address these limitations, this paper explores a practical and challenging setting: ''source-free domain adaptive point cloud completion'', where a well-trained source model must adapt to the target data distribution without access to source data, aiming to improve completion performance. To tackle this problem, we propose a novel method called ''Dual-Stage Preservation and Fusion'' (DSPF), which comprises two key training stages tailored to this new setting. In the source preservation stage, we introduce graph structural alignment and marginal feature alignment to preserve and transfer essential knowledge from the source domain. In the target fusion stage, we design a self-supervised loss to capture the geometric structure of target instances and establish a bidirectional interaction mechanism to transfer partial source knowledge to the target distribution. Extensive experiments on various cross-domain point cloud completion benchmarks demonstrate that our proposed DSPF significantly outperforms existing methods, validating its effectiveness and robustness in source-free domain adaptation scenarios. Our code is available at https://github.com/ZhiXia-SEU/DSPF.

To fully leverage communication resources for pervasive sensing within existing architectures, it is crucial to enable perception through the use of transmitted data. Unlike traditional waveform designs that focus exclusively on modulation, this paper introduces a novel approach starting from the information bit level. By applying controlled perturbations, approximate codewords are aligned with deterministic optimization waveforms, effectively suppressing sidelobe levels and enhancing data-driven sensing capabilities. Additionally, to address decoding impairments caused by these perturbations, we propose a closed-loop reception algorithm that progressively removes residual disturbances while maintaining decoding accuracy. Simulations confirm the algorithm’s effectiveness in balancing communication and sensing performance. Compared to current time-division processing technologies, such as 5G frame structure perception based on reference signals, the proposed waveforms more effectively resolves the trade-off between communication and sensing.

The absence of temporal information in single-temporal satellite remote sensing images presents a substantial challenge for target state discrimination. In this letter, a pioneering Remote Sensing Airplane State Discrimination Network (RSASDNet) is introduced, by leveraging the relationship between targets and their backgrounds in single-temporal high-resolution remote sensing images. To facilitate the study, we take airplane state discrimination as an example, and a Remote Sensing Airport Panoptic Segmentation with Airplane States Dataset (RSAPS-ASD) is constructed. RSASDNet incorporates two key innovations: 1) a scene knowledge graph generation module that constructs scene knowledge representation by capturing spatial relationships between airplane instances and their surrounding environment (e.g., taxiways and hangars); and 2) a novel graph-image hybrid convolution discrimination module that synergistically integrates structural knowledge and spatial semantic information through dedicated dual-branch learning. The effectiveness of the proposed method is validated using RSAPS-ASD, with experimental results demonstrating that RSASDNet achieves an impressive accuracy of 73.95% in airplane state discrimination.

Digital elevation models (DEMs) are crucial geographical data source whereas the resolution of commonly used DEM products is low and cannot meet requirement of some detailed geo-related applications. Deep learning-based methods have been demonstrated to be effective in super-resolution (SR) techniques, which reconstruct high-resolution (HR) images from low-resolution (LR) images. However, existing deep learning methods have not fully considered the multi-scale spatial heterogeneity and topographic knowledge of DEM data that differentiate them from traditional images. These inevitably lead to the localized smoothing of the reconstructed DEM and influence the reliability downstream geographical analysis. This study proposes a transformer-based topography neural network (TTSR) for DEM SR incorporating a local-global deformable block (LGDB) for capturing the multi-scale spatial heterogeneity and topographic knowledge, a spatio-channel coupled channel attention (SimAM) mechanism for reallocating channel weights and providing a supplement of the global spatial features, and an improved terrain loss function (iLoss) for mitigating noise across datasets. TTSR was validated using two publicly available real-world DEM datasets for recovering DEM from 30 m to 10 m. The root mean square error (RMSE) of the proposed method was reduced by approximately 6-30%, 4-16%, and 1-9% in elevation accuracy, slope accuracy, and aspect accuracy, respectively, compared to the best one of those state-of-the-art methods. This research provides new insights for improving the accuracy of the DEM SR, which will help generate global high-resolution terrain products to geographical studies in the future.

Though deep learning methods have achieved the state-of-the-art performance for hyperspectral image (HSI) classification, they often highly rely on large amount of samples for training, and introduce few-shot challenge due to the lack of labeled samples. In this paper, a self-supervised method is presented for HSI classification in the few-shot scenario, where masked autoencoder is employed to reconstruct the masked bands in spectral domain for model pretraining with limited labeled sample, namely Spectral-MAE. The proposed method not only avoids the overfitting via the pretraining, but also provides the model’s ability for effective feature extraction while avoiding the high spatial redundancy. Experiments conducted verify the effectiveness of the proposed method for HSI classification in few-shot situation as compared with other methods.

High-performance target detection algorithms have been proposed for ship detection in Synthetic Aperture Radar (SAR) images in recent years. However, most of them are applied in the situation of single-polarization SAR images and rarely consider the important polarization information in SAR images. In this paper, a polarization-guided strategy is presented to improve the detection performance in single polarization SAR images by predicting polarization type. In which, an extra polarization-guided head is employed following the backbone network to guide the feature maps from the backbone to explore polarization information. Then, the feature map with the polarization information is fused with those from backbone network to enhance feature extraction in feature pyramid networks (FPN), and yields better detection performance. Experiments conducted demonstrate the effectiveness of the proposed method, which can be easily merged with different detectors.

Exploring the relationship between spike phenotypes and wheat yield is crucial for selecting wheat ideotypes, but remains a subject of ongoing debate, primarily due to the lack of efficient spike phenotyping methods, particularly in field environments with complex light conditions. Light detection and ranging (lidar) can precisely capture three-dimensional plant information, minimally affected by light conditions, providing an ideal data source for addressing the abovementioned bottleneck. However, few studies have successfully segmented individual spikes from field-collected lidar data, hindering the extraction of spike phenotypes. Here, we present a novel approach that integrates the kernel-predicting convolution neural network with density-based spatial clustering and Laplacian-based region growing techniques for spike segmentation. Our results showed that the proposed approach enabled accurate segmentation of individual spikes, yielding an F-score of 84.62%. Eight spike phenotypes were successfully extracted from individual spike lidar data, including spike density, spike length, spike width, spike curvature, spike inclination angle, spike height, spike area, and spike volume. Notably, the accuracy of spike length and spike width reached levels of 99% and 65% respectively, with relative root-mean-squared errors of 3.99% and 32.03%. All spike phenotypes exhibited significant positive correlations with wheat yield, collectively accounting for 53% of the variations in wheat yield as determined by a random forest model. The characteristics of spike phenotypes were effective indicators for discerning yield variations among wheat varieties, highlighting spike phenotypes hold significant value in wheat ideotype selection, and lidar has great potential to expedite the field-based wheat breeding cycle.

In this paper, we present the statistical characteristics of Non-Gaussian sea surfaces, and find suitable statistical indicators to distinguish the rough surfaces generated by different parameters. Firstly, according to the sea surface height and wave spectrum, the rough sea surface is modeled. Then the non-Gaussian rough sea surface is constructed by using the SAR geometry. Finally, the roughness index is generated according to the SAR imaging results. The method can allow more degrees of freedom when generating a non-Gaussian rough sea surface, rather than just specifying the difference of skewness and kurtosis. Results analysis verifies the effectiveness of the proposed method.

Real-time positioning of flying aircraft based on hyperspectral remote sensing images is a novel application of satellite remote sensing. The complex aviation environment proposes huge challenges to accurately obtain features of flying aircraft from hyperspectral images (HSI) based on the traditional remote sensing image processing method. This paper proposes an automatic detection method for flying aircraft by Gaofen-5 (GF-5) HSI. This study firstly acquires the candidate target regions by detecting anomalies among the spectral inter-bands based on the imaging characteristics of the GF-5 remote sensing satellite sensor and the features of flying aircraft in the remote sensing image. The flying aircraft is then detected by confirming its features of a linear combination and proportional displacement. The experimental results demonstrate that the detection accuracy of flying aircraft can reach 99.9% under good weather conditions. Furthermore, the proposed method can effectively reduce the computational cost and realize the real-time detection of flying aircraft, which can be used to detect flying targets in large-scale HSI.

With the rapid development of high throughput satellite and low-earth-orbit (LEO) satellite communication, the electronically steered array (ESA) antennas are desired. This paper has presented the design of a low profile Ka-band 256-element transmit dual-circularly-polarized planar phased array, which consists of $16 \times 16$ radiating antenna elements, 64 8-channel beamformers, a beam-forming network and a beam control unit. This antenna system has the advantages of low profile, light weight, high integration and flexible extensibility. It is a promising solution for LEO satcom application.

This paper has presented the design of a low profile K-band 256-Element receive dual-circularly-polarized planar phased array, which consists of $16 \times 16$ radiating antenna elements, 64 beamformer chips, a beam-forming network and a beam control unit. This antenna system has the advantages of low profile, light weight, high integration and flexible extensibility. It is a promising solution for LEO satcom application.

Forest inventory holds an essential role in forest management and research, but the existing field inventory methods are highly time-consuming and labor-intensive. Here, we developed a simultaneous localization and mapping-based backpack light detection and ranging (LiDAR) system with dual orthogonal laser scanners and an open-source Python package called Forest3D for efficient and accurate forest inventory applications. Two key forest inventory variables, tree height and diameter at breast height (DBH), were extracted at six study sites with different tree species compositions. In addition, the vertical point density distribution and leaf area density (LAD) were calculated for two complex natural forest sites. The results showed that the backpack LiDAR system together with the Forest3D package accurately estimated the tree height ( R2= 0.65, RMSE = 1.90 m) and DBH ( R2= 0.95, RMSE = 0.02 m), which were equivalent to those derived from terrestrial laser scanning (TLS), but with much higher efficiency. The point density of the backpack LiDAR data was higher than or the same as that of the TLS data across all height strata, and the estimated LAD fit well with the TLS estimates ( R2> 0.92, RMSE = 0.01 m2/m3). The backpack LiDAR system, along with the Forest3D package, provides an efficient and accurate solution for extracting forest inventory variables, which should be of great interests to forest managers and researchers.

The emerging near-surface light detection and ranging (LiDAR) platforms [e.g., terrestrial, backpack, mobile, and unmanned aerial vehicle (UAV)] have shown great potential for forest inventory. However, different LiDAR platforms have limitations either in data coverage or in capturing undercanopy information. The fusion of multiplatform LiDAR data is a potential solution to this problem. Because of the complexity and irregularity of forests and the inaccurate positioning information under forest canopies, current multiplatform data fusion still involves substantial manual efforts. In this article, we proposed an automatic multiplatform LiDAR data registration framework based on the assumption that each forest has a unique tree distribution pattern. Five steps are included in the proposed framework, i.e., individual tree segmentation, triangulated irregular network (TIN) generation, TIN matching, coarse registration, and fine registration. TIN matching, as the essential step to find the corresponding tree pairs from multiplatform LiDAR data, uses a voting strategy based on the similarity of triangles composed of individual tree locations. The proposed framework was validated by fusing backpack and UAV LiDAR data and fusing multiscan terrestrial LiDAR data in coniferous forests. The results showed that both registration experiments could reach a satisfying data registration accuracy (horizontal root-mean-square error (RMSE) <; 30 cm and vertical RMSE <; 20 cm). Moreover, the proposed framework was insensitive to individual tree segmentation errors, when the individual tree segmentation accuracy was higher than 80%. We believe that the proposed framework has the potential to increase the efficiency of accurately registering multiplatform LiDAR data in forest environments.

Separating structural components is important but also challenging for plant phenotyping and precision agriculture. Light detection and ranging (LiDAR) technology can potentially overcome these difficulties by providing high quality data. However, there are difficulties in automatically classifying and segmenting components of interest. Deep learning can extract complex features, but it is mostly used with images. Here, we propose a voxel-based convolutional neural network (VCNN) for maize stem and leaf classification and segmentation. Maize plants at three different growth stages were scanned with a terrestrial LiDAR and the voxelized LiDAR data were used as inputs. A total of 3000 individual plants (22 004 leaves and 3000 stems) were prepared for training through data augmentation, and 103 maize plants were used to evaluate the accuracy of classification and segmentation at both instance and point levels. The VCNN was compared with traditional clustering methods (K-means and density-based spatial clustering of applications with noise), a geometry-based segmentation method, and state-of-the-art deep learning methods (PointNet and PointNet++). The results showed that: 1) at the instance level, the mean accuracy of classification and segmentation (F-score) were 1.00 and 0.96, respectively; 2) at the point level, the mean accuracy of classification and segmentation (F-score) were 0.91 and 0.89, respectively; 3) the VCNN method outperformed traditional clustering methods; and 4) the VCNN was on par with PointNet and PointNet++ in classification, and performed the best in segmentation. The proposed method demonstrated LiDAR's ability to separate structural components for crop phenotyping using deep learning, which can be useful for other fields.

Accurate and high throughput extraction of crop phenotypic traits, as a crucial step of molecular breeding, is of great importance for yield increasing. However, automatic stem-leaf segmentation as a prerequisite of many precise phenotypic trait extractions is still a big challenge. Current works focus on the study of the 2-D image-based segmentation, which are sensitive to illumination and occlusion. Light detection and ranging (LiDAR) can obtain accurate 3-D information with its active laser scanning and strong penetration ability, which breaks through phenotyping from 2-D to 3-D. However, few researches have addressed the problem of the LiDAR-based stem-leaf segmentation. In this paper, we proposed a median normalized-vector growth (MNVG) algorithm, which can segment stem and leaf with four steps, i.e., preprocessing, stem growth, leaf growth, and postprocessing. The MNVG method was tested by 30 maize samples with different heights, compactness, leaf numbers, and densities from three growing stages. Moreover, phenotypic traits at leaf, stem, and individual levels were extracted with the truly segmented instances. The mean accuracy of segmentation at point level in terms of the recall, precision, F-score, and overall accuracy were 0.92, 0.93, 0.92, and 0.93, respectively. The accuracy of phenotypic trait extraction in leaf, stem, and individual levels ranged from 0.81 to 0.95, 0.64 to 0.97, and 0.96 to 1, respectively. To our knowledge, this paper proposed the first LiDAR-based stem-leaf segmentation and phenotypic trait extraction method in agriculture field, which may contribute to the study of LiDAR-based plant phonemics and precise agriculture.

Smart electricity meters have been replacing conventional meters worldwide, enabling automated collection of fine-grained (e.g., every 15 minutes or hourly) consumption data. A variety of smart meter analytics algorithms and applications have been proposed, mainly in the smart grid literature. However, the focus has been on what can be done with the data rather than how to do it efficiently. In this article, we examine smart meter analytics from a software performance perspective. First, we design a performance benchmark that includes common smart meter analytics tasks. These include offline feature extraction and model building as well as a framework for online anomaly detection that we propose. Second, since obtaining real smart meter data is difficult due to privacy issues, we present an algorithm for generating large realistic datasets from a small seed of real data. Third, we implement the proposed benchmark using five representative platforms: a traditional numeric computing platform (Matlab), a relational DBMS with a built-in machine learning toolkit (PostgreSQL/MADlib), a main-memory column store (“System C”), and two distributed data processing platforms (Hive and Spark/Spark Streaming). We compare the five platforms in terms of application development effort and performance on a multicore machine as well as a cluster of 16 commodity servers.