Cloud Features Research Articles

Receptive Field Space for Point Cloud Analysis.

Similar to convolutional neural networks for image processing, existing analysis methods for 3D point clouds often require the designation of a local neighborhood to describe the local features of the point cloud. This local neighborhood is typically manually specified, which makes it impossible for the network to dynamically adjust the receptive field's range. If the range is too large, it tends to overlook local details, and if it is too small, it cannot establish global dependencies. To address this issue, we introduce in this paper a new concept: receptive field space (RFS). With a minor computational cost, we extract features from multiple consecutive receptive field ranges to form this new receptive field space. On this basis, we further propose a receptive field space attention mechanism, enabling the network to adaptively select the most effective receptive field range from RFS, thus equipping the network with the ability to adjust granularity adaptively. Our approach achieved state-of-the-art performance in both point cloud classification, with an overall accuracy (OA) of 94.2%, and part segmentation, achieving an mIoU of 86.0%, demonstrating the effectiveness of our method.

A simulation‐assisted point cloud segmentation neural network for human–robot interaction applications

AbstractWith the advancement of industrial automation, the frequency of human–robot interaction (HRI) has significantly increased, necessitating a paramount focus on ensuring human safety throughout this process. This paper proposes a simulation‐assisted neural network for point cloud segmentation in HRI, specifically distinguishing humans from various surrounding objects. During HRI, readily accessible prior information, such as the positions of background objects and the robot's posture, can generate a simulated point cloud and assist in point cloud segmentation. The simulation‐assisted neural network utilizes simulated and actual point clouds as dual inputs. A simulation‐assisted edge convolution module in the network facilitates the combination of features from the actual and simulated point clouds, updating the features of the actual point cloud to incorporate simulation information. Experiments of point cloud segmentation in industrial environments verify the efficacy of the proposed method.

Perceptual point cloud qality assessment for immersive metaverse experience

Perceptual quality assessment for point cloud is critical for immersive metaverse experience and is a challenging task. Firstly, because point cloud is formed by unstructured 3D points that makes the topology more complex. Second, the quality impairment generally involves both geometric attributes and color properties, where the measurement of the geometric distortion becomes more complex. We proposed a perceptual point cloud quality assessment model that follows the perceptual features of Human Visual System (HVS) and the intrinsic characteristics of the point cloud. The point cloud is first pre-processed to extract the geometric skeleton keypoints with graph filtering-based re-sampling, and local neighboring regions around the geometric skeleton keypoints are constructed by K-Nearest Neighbors (KNN) clustering. For geometric distortion, the Point Feature Histogram (PFH) is extracted as the feature descriptor, and the Earth Mover's Distance (EMD) between the PFHs of the corresponding local neighboring regions in the reference and the distorted point clouds is calculated as the geometric quality measurement. For color distortion, the statistical moments between the corresponding local neighboring regions are computed as the color quality measurement. Finally, the global perceptual quality assessment model is obtained as the linear weighting aggregation of the geometric and color quality measurement. The experimental results on extensive datasets show that the proposed method achieves the leading performance as compared to the state-of-the-art methods with less computing time. Meanwhile, the experimental results also demonstrate the robustness of the proposed method across various distortion types. The source codes are available at https://github.com/llsurreal919/PointCloudQualityAssessment.

Integrating PointNet-Based Model and Machine Learning Algorithms for Classification of Rupture Status of IAs.

The rupture of intracranial aneurysms (IAs) would result in subarachnoid hemorrhage with high mortality and disability. Predicting the risk of IAs rupture remains a challenge. This paper proposed an effective method for classifying IAs rupture status by integrating a PointNet-based model and machine learning algorithms. First, medical image segmentation and reconstruction algorithms were applied to 3D Digital Subtraction Angiography (DSA) imaging data to construct three-dimensional IAs geometric models. Geometrical parameters of IAs were then acquired using Geomagic, followed by the computation of hemodynamic clouds and hemodynamic parameters using Computational Fluid Dynamics (CFD). A PointNet-based model was developed to extract different dimensional hemodynamic cloud features. Finally, five types of machine learning algorithms were applied on geometrical parameters, hemodynamic parameters, and hemodynamic cloud features to classify and recognize IAs rupture status. The classification performance of different dimensional hemodynamic cloud features was also compared. The 16-, 32-, 64-, and 1024-dimensional hemodynamic cloud features were extracted with the PointNet-based model, respectively, and the four types of cloud features in combination with the geometrical parameters and hemodynamic parameters were respectively applied to classify the rupture status of IAs. The best classification outcomes were achieved in the case of 16-dimensional hemodynamic cloud features, the accuracy of XGBoost, CatBoost, SVM, LightGBM, and LR algorithms was 0.887, 0.857, 0.854, 0.857, and 0.908, respectively, and the AUCs were 0.917, 0.934, 0.946, 0.920, and 0.944. In contrast, when only utilizing geometrical parameters and hemodynamic parameters, the accuracies were 0.836, 0.816, 0.826, 0.832, and 0.885, respectively, with AUC values of 0.908, 0.922, 0.930, 0.884, and 0.921. In this paper, classification models for IAs rupture status were constructed by integrating a PointNet-based model and machine learning algorithms. Experiments demonstrated that hemodynamic cloud features had a certain contribution weight to the classification of IAs rupture status. When 16-dimensional hemodynamic cloud features were added to the morphological and hemodynamic features, the models achieved the highest classification accuracies and AUCs. Our models and algorithms would provide valuable insights for the clinical diagnosis and treatment of IAs.

Deep learning-based prediction of plant height and crown area of vegetable crops using LiDAR point cloud

Remote sensing has been increasingly used in precision agriculture. Buoyed by the developments in the miniaturization of sensors and platforms, contemporary remote sensing offers data at resolutions finer enough to respond to within-farm variations. LiDAR point cloud, offers features amenable to modelling structural parameters of crops. Early prediction of crop growth parameters helps farmers and other stakeholders dynamically manage farming activities. The objective of this work is the development and application of a deep learning framework to predict plant-level crop height and crown area at different growth stages for vegetable crops. LiDAR point clouds were acquired using a terrestrial laser scanner on five dates during the growth cycles of tomato, eggplant and cabbage on the experimental research farms of the University of Agricultural Sciences, Bengaluru, India. We implemented a hybrid deep learning framework combining distinct features of long-term short memory (LSTM) and Gated Recurrent Unit (GRU) for the predictions of plant height and crown area. The predictions are validated with reference ground truth measurements. These predictions were validated against ground truth measurements. The findings demonstrate that plant-level structural parameters can be predicted well ahead of crop growth stages with around 80% accuracy. Notably, the LSTM and the GRU models exhibited limitations in capturing variations in structural parameters. Conversely, the hybrid model offered significantly improved predictions, particularly for crown area, with error rates for height prediction ranging from 5 to 12%, with deviations exhibiting a more balanced distribution between overestimation and underestimation This approach effectively captured the inherent temporal growth pattern of the crops, highlighting the potential of deep learning for precision agriculture applications. However, the prediction quality is relatively low at the advanced growth stage, closer to the harvest. In contrast, the prediction quality is stable across the three different crops. The results indicate the presence of a robust relationship between the features of the LiDAR point cloud and the auto-feature map of the deep learning methods adapted for plant-level crop structural characterization. This approach effectively captured the inherent temporal growth pattern of the crops, highlighting the potential of deep learning for precision agriculture applications.

Explosion characteristics of AlH3 dust cloud with varying micron particle sizes

The explosion characteristics of AlH3 dust clouds with varying micron particle sizes are investigated in this paper. The explosion mechanism of AlH3 dust clouds is initially explored. The results indicate that the activation energy for hydrogen evolution shows an increasing trend with the AlH3 particle size, and both of these values are comparable to the 30 kJ/mol desorption energy of hydrogen from large AlH3 clusters. Due to the rapid dehydrogenation rate and small-size residual aluminum core, fine AlH3 particles exhibit a continuous gas-phase flame and fast propagation velocity. With increasing particle size, the lower dehydrogenation rate, reduced specific surface area, and higher volume heat capacity contribute to a higher level of incomplete reaction. Consequently, large micro-sized AlH3 particles display a discrete flame, slower flame velocity, and lower explosion pressure. The flame delamination phenomenon is not observed in our experiments on AlH3 dust clouds.

Change in precipitation pattern over South Asia in response to the trends in regional warming and free-tropospheric aerosol loading

Spatial and temporal shifts in rainfall patterns over South Asia and the adjoining Seas during the pre-monsoon season have been observed over the past 2 decades from 2000 to 2019. Aerosol particles suspended above the boundary layer are a contributing factor to these changes. These particles not only alter cloud characteristics, but also diminish the lapse rate, thereby suppressing convective activity, leading to precipitation anomalies. Over the past 2 decades, high-rainfall regions have experienced declining rainfall, while low-rainfall regions have received increased rainfall. Coinciding with notable anomalies in precipitation, contrasting trends in aerosol optical depth, particularly due to absorbing aerosols in the elevated regions of the atmosphere, are seen. Apart from aerosols, several factors are considered that are critical in modifying precipitation patterns over the study region, such as water vapor content, convective processes, and lower-level relative humidity. We observed a potential transport of excess water vapor by ambient circulation from the oceanic regions having reduced rain, such as Bay of Bengal and the Arabian Sea, to higher latitudes enabling precipitation anomaly at distant locations.

Application of multi-source data fusion on intelligent prediction of photovoltaic power

Accurate photovoltaic power prediction is crucial for reducing the uncertainty of photovoltaic systems in grid operation. Variations in cloud cover are one of the primary factors leading to fluctuations in photovoltaic power generation. However, clouds possess highly complex three-dimensional structures. Existing photovoltaic power prediction methods typically rely on two-dimensional cloud images, which are insufficient for fully capturing the impact of clouds on photovoltaic power generation. To address these challenges, this paper proposes a Multi-source data Photovoltaic power Prediction Model (MPPM) based on satellite cloud images and meteorological data. MPPM mainly includes SpatioTemporal feature Conditional Diffusion Model (STCDM), Attention Stacked LSTM network (ASLSTM) and Multidimensional Feature Fusion Module (MFFM). STCDM model utilizes a diffusion model to accurately forecast two-dimensional satellite cloud imagery. ASLSTM extracts the features of three-dimensional meteorological elements. MFFM module integrates the two-dimensional satellite cloud imagery features with the three-dimensional meteorological element features. The two-dimensional satellite cloud imagery reflects the visible aspect of cloud layers, while the three-dimensional meteorological data, which includes cloud water and ice content correlated with altitude, mainly captures the invisible aspect of cloud layers. These two sets of information complement each other, enabling a more comprehensive capture of the three-dimensional characteristics of clouds. Satellite cloud image prediction experiment and photovoltaic power prediction experiment are carried out on STCDM and MPPM model respectively. The experimental results demonstrate that the STCDM model achieved a Structural Similarity index (SSIM) of 0.909 for predicting satellite cloud imagery within 1 h and 0.789 within 24 h. Meanwhile, the MPPM model attained a pearson Correlation coefficient (CORR) of 0.945 for predicting PV power within 1 h and 0.856 within 24 h. These findings indicate that both the STCDM and MPPM models outperform other comparative algorithms in forecasting satellite cloud imagery and PV power.

On the characterization of Cloud occurrence and its impact on solar radiation in Mbour, Senegal

The objective of this study is to evaluate the clouds seasonal occurrence characteristics, and to estimate their impact on solar radiation in Mbour, Senegal, West Africa. Here, we use datasets from various sources including: i) observations from the Clouds and Earth's Radiant Energy System satellite sensors, ii) in situ shortwave radiation measurement obtained from the Mbour station, and iii) the outgoing longwave radiation (OLR) obtained from the National Centers for Environmental Prediction reanalysis data. Results show a marked seasonality, associated with high spatial variation in terms of cloud occurrence over Senegal. The maximum cloud occurrences are observed during the wet summer season (June–October), whilst the minimum cloud occurrences are recorded during the long-dry season from November to May. During the monsoon season the cloud activity becomes more intense with a total cloud cover of about 80%, a cloud optical depth of around 7, and a high convective activity illustrated by a low OLR (below 240 W/m2). Likewise, across Senegal a strong north-south gradient of the cloud characteristics is observed. Based on quantitative comparison between cloud occurrence and radiation measurement, results show an important seasonal impact on available solar potential in Mbour. Conversely to the cloud occurrence, the maximum of both direct normal and global solar potentials is recorded during the dry season, coinciding with the period with clean sky. An investigation of the cloud influence on solar radiation on selected study cases indicates a decrease of 60% (80%) for the total (direct normal) radiation during the peak of the summer monsoon season.

Using the full potential of Airborne Laser Scanning (aerial LiDAR) in wildlife research

AbstractSpecies' habitats are strongly influenced by the 3‐dimensional (3D) structure of ecosystems. The dominant technique used to measure 3D structure is Airborne Laser Scanning (ALS), a type of LiDAR (Light Detection and Ranging) technology. Airborne Laser Scanning captures fine‐scale structural information over large spatial extents and provides useful environmental predictors for habitat modeling. However, due to technical complexities of processing ALS data, the full potential of ALS is not yet realized in wildlife research, with most studies relying on a limited set of 3D predictors, such as vegetation metrics developed principally for forestry applications. Here, we highlight the full potential of ALS data for wildlife research and provide insight into how it can be best used to capture the environmental conditions, resources, and risks that directly determine a species' habitat. We provide a nontechnical overview of ALS data, covering data considerations and the modern options available for creating custom, ecologically relevant, ALS predictors. Options included the following: i) direct point cloud approaches that measure structure using grid, voxel, and point metrics, ii) object‐based approaches that identify user‐defined features in the point cloud, and iii) modeled environmental predictors that use additional modeling to infer a range of habitat characteristics, including the extrapolation of field acquired measurements over ALS data. By using custom ALS predictors that capture species‐specific resources, risks, and environmental conditions, wildlife practitioners can produce models that are tailored to a species' ecology, have greater biological realism, test a wider range of species‐environment relationships across scales, and provide more meaningful insights to inform wildlife conservation and management.

FF3D: A Rapid and Accurate 3D Fruit Detector for Robotic Harvesting.

This study presents the Fast Fruit 3D Detector (FF3D), a novel framework that contains a 3D neural network for fruit detection and an anisotropic Gaussian-based next-best view estimator. The proposed one-stage 3D detector, which utilizes an end-to-end 3D detection network, shows superior accuracy and robustness compared to traditional 2D methods. The core of the FF3D is a 3D object detection network based on a 3D convolutional neural network (3D CNN) followed by an anisotropic Gaussian-based next-best view estimation module. The innovative architecture combines point cloud feature extraction and object detection tasks, achieving accurate real-time fruit localization. The model is trained on a large-scale 3D fruit dataset and contains data collected from an apple orchard. Additionally, the proposed next-best view estimator improves accuracy and lowers the collision risk for grasping. Thorough assessments on the test set and in a simulated environment validate the efficacy of our FF3D. The experimental results show an AP of 76.3%, an AR of 92.3%, and an average Euclidean distance error of less than 6.2 mm, highlighting the framework's potential to overcome challenges in orchard environments.

LiDAR point cloud simplification algorithm with fuzzy encoding-decoding mechanism

With the explosive growth in the density of acquired point cloud data, point cloud processing tasks will face tremendous challenges. LiDAR point cloud simplification is a key phase in addressing this issue, which effectively promotes the development of LiDAR technology in many engineering fields. In this study, an innovative point cloud simplification algorithm with the fuzzy encoding-decoding mechanism is proposed. In the developed scheme, an approach for curvature estimation is first designed on the basis of the k-neighbor searching and principal component analysis. Then, a collection of feature point sets is set up with the ordered curvatures. Subsequently, a Fuzzy C-Means clustering based encoding mechanism is employed to capture the level point cloud structures in depth and establish a reasonable and streamlined strategy for point clouds. Each feature point set and non-feature point set are encoded into a prototype matrix and a partition (membership) matrix. The membership degree of each feature point to its prototype becomes the basis for the simplification strategy. Finally, the simplification result of the point cloud is formed through merging the simplification results of all subsets. The method proposed in this study effectively preserves the point cloud features and ensures a uniform distribution of the simplified point cloud. A comparative analysis of the point cloud simplification is conducted. The experimental results demonstrate that the developed algorithm outperformed other point cloud simplification algorithms.

Bridging the Domain Gap in Scene Flow Estimation via Hierarchical Smoothness Refinement

This article introduces SmoothFlowNet3D, an innovative encoder-decoder architecture specifically designed for bridging the domain gap in scene flow estimation. To achieve this goal, SmoothFlowNet3D divides the scene flow estimation task into two stages: initial scene flow estimation and smoothness refinement. Specifically, SmoothFlowNet3D comprises a hierarchical encoder that extracts multi-scale point cloud features from two consecutive frames, along with a hierarchical decoder responsible for predicting the initial scene flow and further refining it to achieve smoother estimation. To generate the initial scene flow, a cross-frame nearest-neighbor search operation is performed between the features extracted from two consecutive frames, resulting in forward and backward flow embeddings. These embeddings are then combined to form the bidirectional flow embedding, serving as input for predicting the initial scene flow. Additionally, a flow smoothing module based on the self-attention mechanism is proposed to predict the smoothing error and facilitate the refinement of the initial scene flow for more accurate and smoother estimation results. Extensive experiments demonstrate that the proposed SmoothFlowNet3D approach achieves state-of-the-art performance on both synthetic datasets and real LiDAR point clouds, confirming its effectiveness in enhancing scene flow smoothness.

An End-to-End Geometric Characterization-aware Semantic Instance Segmentation Network for ALS Point Clouds

Abstract. Semantic instance segmentation from scenes, serving as a crucial role for 3D modelling and scene understanding. Conducting semantic segmentation before grouping instances is adopted by the existing state-of-the-art methods. However, without additional refinement, semantic errors will fully propagate into the grouping stage, resulting in low overlap with the ground truth instance. Furthermore, the proposed methods focused on indoor level scenes, which are limited when directly applied to large-scale outdoor Airborne Laser Scanning (ALS) point clouds. Numerous instances, significant object density and scale variations make ALS point clouds distinct from indoor data. In order to address the problems, we proposed a geometric characterization-aware semantic instance segmentation network, which utilized both semantic and objectness score to select potential points for grouping. And in point cloud feature learning stage, hand-craft geometry features are taken as input for geometric characterization awareness. Moreover, to address errors propagated from previous modules after grouping, we have additionally designed a per-instance refinement module. To assess semantic instance segmentation, we conducted experiments on an open-source dataset. Additionally, we performed semantic segmentation experiments to evaluate the performance of our proposed point cloud feature learning method.

A selective harvesting robot for cherry tomatoes: Design, development, field evaluation analysis

AbstractWith the aging population and increasing labor costs, traditional manual harvesting methods have become less economically efficient. Consequently, research into fully automated harvesting using selective harvesting robots for cherry tomatoes has become a hot topic. However, most of the current research is focused on individual harvesting of large tomatoes, and there is less research on the development of complete systems for harvesting cherry tomatoes in clusters. The purpose of this study is to develop a harvesting robot system capable of picking tomato clusters by cutting their fruit‐bearing pedicels and to evaluate the robot prototype in real greenhouse environments. First, to enhance the grasping stability, a novel end‐effector was designed. This end‐effector utilizes a cam mechanism to achieve asynchronous actions of cutting and grasping with only one power source. Subsequently, a visual perception system was developed to locate the cutting points of the pedicels. This system is divided into two parts: rough positioning of the fruits in the far‐range view and accurate positioning of the cutting points of the pedicels in the close‐range view. Furthermore, it possesses the capability to adaptively infer the approaching pose of the end‐effector based on point cloud features extracted from fruit‐bearing pedicels and stems. Finally, a prototype of the tomato‐harvesting robot was assembled for field trials. The test results demonstrate that in tomato clusters with unobstructed pedicels, the localization success rates for the cutting points were 88.5% and 83.7% in the two greenhouses, respectively, while the harvesting success rates reached 57.7% and 55.4%, respectively. The average cycle time to harvest a tomato cluster was 24 s. The experimental results prove the potential for commercial application of the developed tomato‐harvesting robot and through the analysis of failure cases, discuss directions for future work.

The puzzling north polar region of Uranus: Continued zero-shear winds and increasing brightness from 2015 through 2022 according to 7 years of Keck AO imaging

There may be a permanent north–south asymmetry in the zonal wind profile of Uranus. According to high signal-to-noise near-IR observations with the Keck telescope NIRC2 camera, through 2014 the winds north of 60∘N had a constant (zero shear) longitudinal drift rate of 4.1∘/h westward (Sromovsky et al., 2015). Yet the south polar winds in 1986 had strong latitudinal shear, reaching speeds of 8.5∘/h (Karkoschka, 2015). If that asymmetry were a result of extreme seasonal forcing, it might be expected to reverse before the 2030 solstice. However, continued high signal-to-noise near-IR observations from 2015 through 2022 provide little evidence that the north polar winds have changed at all, although there have been changes in the distribution of cloud features used to track those winds and an increase in the brightness of the background haze that makes it harder to detect these small low-contrast features. Between 2015 and 2017, a gap in easily trackable features developed in the 55∘N–60∘N latitude region, and at the northern edge of that gap there was a hint that a dramatic change to the inverted profile might be starting. To further investigate that possible change and to fill in the gap, we developed a new wind measurement technique that was more sensitive to low-contrast spatial structure. Our new method does not confirm any significant wind speed change in the gap, nor at its northern edge.The lifetimes of many of the small discrete polar features are surprisingly long. Many were tracked over time spans over 24 h, and some lasted for weeks. More prominent features seen at low latitudes seem to persist for multiple years and vary in latitude and drift rate in a cyclical fashion. While measurements so far favor a permanent asymmetry in the zonal wind profile of Uranus, the possibility remains that a dramatic circulation change might take place between 2022 and the next solstice in 2030. Using simplified radiation transfer modeling we infer a largely unchanging upper layer of small particles of very low optical depth vertically distributed above an active deeper mid-level layer, at an effective pressure of ∼2–2.5 bar. That middle layer increased in reflectivity at 1.6 μm by more than a factor of 2 from 2015 to 2022 and is the main source of the top-of-atmosphere brightness increase seen over that period, rather than any significant temporal change in polar methane abundance.

Characteristics of Cloud and Aerosol Derived from Lidar Observations during Winter in Lhasa, Tibetan Plateau

In order to investigate the variations of cloud and aerosol vertical profiles over the Tibetan Plateau (TP) in winter, we performed ground-based lidar observations in Lhasa, a city on the TP, from November 2021 to January 2022. The profiles of extinction coefficient, depolarization ratio, and signal-to-noise ratio (SNR) were retrieved using the atmospheric echo signals collected by the lidar. Clouds were identified by the range-correction echo signals and classified into water clouds, mixed clouds, horizontally oriented ice crystal clouds (HOICC), and ice clouds by the depolarization ratio and the hourly temperature from the European Centre for Medium-Range Weather Forecasts Reanalysis version 5 (ERA5). The clouds mainly appeared at a height of 3~5 km from 14:00–22:00 Beijing Time throughout the field campaign. The height and frequency (~30%) for cloud appearance were significantly lower than that reported in previous studies in summer. The cloud categories were dominated by mixed clouds and ice clouds during the observation period. The proportions of ice clouds gradually increased with increasing heights. After eliminating profiles influenced by clouds, the aerosol extinction coefficient and depolarization ratio were obtained, and the atmospheric boundary layer height (ABLH) was calculated. The aerosol extinction coefficient decreased with increasing height in the ABLH, and there were no obvious changes for the aerosol extinction coefficient above the ABL. The aerosol extinction coefficients near the Earth’s surface presented two peaks, appearing in the morning and evening, respectively. The high aerosols at the surface in the morning continually spread upward for 4–5 h and finally reached an altitude of 1 km with the development of ABLH. In addition, the depolarization ratio of aerosols decreased slowly with increasing altitudes. There was no obvious diurnal variation for depolarization ratios, indicating partly that the source of aerosols did not change significantly. These results are beneficial in understanding the evolution of cloud and aerosol vertical profiles over the TP.

MMFG: Multimodal-based Mutual Feature Gating 3D Object Detection

To address the problem that image and point cloud features are fused in a coarse fusion way and cannot achieve deep fusion, this paper proposes a multimodal 3D object detection architecture based on a mutual feature gating mechanism. First, since the feature aggregation approach based on the set abstraction layer cannot obtain fine-grained features, a point-based self-attention mechanism module is designed. This module is added to the extraction branch of point cloud features to achieve fine-grained feature aggregation while maintaining accurate location information. Second, a new gating mechanism is designed for the deep fusion of image and point cloud. Deep fusion is achieved by mutual feature weighting between the image and the point cloud. The newly fused features are then fed into a feature refinement network to extract classification confidence and 3D target bounding boxes. Finally, a multi-scale detection architecture is proposed to obtain a more complete object shape. The location-based encoding feature algorithm is also designed to focus the interest points in the region of interest adaptively. The whole architecture shows outstanding performance on the KITTI3D and nuSenece datasets, especially at the difficult level. It shows that the framework solves the problem of low detection rates in LiDAR mode due to the low number of surface points obtained from distant objects.

Radiation Feature Fusion Dual-Attention Cloud Segmentation Network

In the field of remote sensing image analysis, the issue of cloud interference in high-resolution images has always been a challenging problem, with traditional methods often facing limitations in addressing this challenge. To this end, this study proposes an innovative solution by integrating radiative feature analysis with cutting-edge deep learning technologies, developing a refined cloud segmentation method. The core innovation lies in the development of FFASPPDANet (Feature Fusion Atrous Spatial Pyramid Pooling Dual Attention Network), a feature fusion dual attention network improved through atrous spatial convolution pooling to enhance the model’s ability to recognize cloud features. Moreover, we introduce a probabilistic thresholding method based on pixel radiation spectrum fusion, further improving the accuracy and reliability of cloud segmentation, resulting in the “FFASPPDANet+” algorithm. Experimental validation shows that FFASPPDANet+ performs exceptionally well in various complex scenarios, achieving a 99.27% accuracy rate in water bodies, a 96.79% accuracy rate in complex urban settings, and a 95.82% accuracy rate in a random test set. This research not only enhances the efficiency and accuracy of cloud segmentation in high-resolution remote sensing images but also provides a new direction and application example for the integration of deep learning with radiative algorithms.

Retrieved Atmospheres and Inferred Surface Properties for Terrestrial Exoplanets Using Transmission and Reflected-light Spectroscopy

Future astrophysics missions will seek extraterrestrial life via transmission and direct-imaging observations. To assess habitability and biosignatures, we need robust retrieval tools to analyze observed spectra, and infer surface and atmospheric properties with their uncertainties. We use a novel retrieval tool to assess accuracy in characterizing near-surface habitability and biosignatures via simulated transmission and direct-imaging spectra, based on the Origins Space Telescope (Origins) and LUVOIR mission concepts. We assess our ability to discriminate between an Earth-like and a false-positive O3 TRAPPIST-1 e with transmission spectroscopy. In reflected light, we assess the robustness of retrieval results to unmodeled cloud extinction. We find that assessing habitability using transmission spectra may be challenging due to relative insensitivity to surface temperature and near-surface H2O abundances. Nonetheless, our order-of-magnitude H2O constraints can discriminate extremely desiccated worlds. Direct imaging is insensitive to surface temperature and subject to the radius/albedo degeneracy, but this method proves highly sensitive to surface water abundance, achieving retrieval precision within 0.1% even with partial clouds. Concerning biosignatures, Origins-like transmission observations (t = 40 hr) may detect the CO2/CH4 pair on M-dwarf planets and differentiate between biological and false-positive O3 using H2O and abundant CO. In contrast, direct-imaging observations with LUVOIR-A (t = 10 hr) are better suited to constraining O2 and O3, and may be sensitive to wavelength-dependent water cloud features, but will struggle to detect modern-Earth-like abundances of methane. For direct imaging, we weakly detect a stratospheric ozone bulge by fitting the near-UV wings of the Hartley band.