The photon point clouds acquired by ICESat-2/ATLAS offer unprecedented potential for Earth observation but are heavily contaminated by noise photons, posing a significant challenge for downstream applications. Traditional denoising methods, which often rely on local density statistics, struggle with complex terrains and varying signal-to-noise ratios. While deep learning presents a promising alternative, existing approaches often inefficiently process the inherently sparse data via 2D projections or non-optimized 3D networks. To address these limitations, this paper introduces a novel deep learning framework for ICESat-2 photon denoising, termed Sparse Manifold Retrieval Network (SMRNet). We propose a Manifold-Aware Convolution (MAC) module to capture the continuous manifold structures of signal photons through multi-scale dilated sparse convolutions, and a Cross-Scale Pyramid Enhancement (CSPE) module to effectively refine multi-level features extracted from the encoder. Evaluated on a manually annotated dataset covering southeastern coastal regions of China, SMRNet demonstrates superior performance over traditional denoising method and data-driven baselines across multiple metrics. The results underscore the effectiveness of SMRNet in enhancing denoising accuracy, particularly in challenging environments with sparse signals and rugged topography.

The growth of distributed energy resources and local energy markets heightens the need for price formation that is transparent, privacy preserving, and compatible with network constraints. Blockchain provides a trust-minimized substrate for auditable clearing and settlement through consensus, tamperevident ledgers, and smart contracts. This survey organizes blockchain-enabled pricing into three families, namely auction-based, game-theoretic, and optimization-based, and links them to enabling techniques such as metering oracles, secure multiparty computation, zero-knowledge proofs, and verifiable optimality certificates. Applications span wholesale electricity, carbon and green certificates, distributed energy trading, ancillary services, and electric vehicles. Evidence indicates gains in auditability, privacy, network awareness, and automated settlement, alongside challenges in scalability, data protection, grid integration, and regulation. The survey distills design patterns and research directions toward verifiable, interoperable, and governable pricing modules that complement system-operator markets.

Airdrops are a widely used mechanism in Web3 ecosystems to incentivize early users by distributing governance tokens. However, these mechanisms are increasingly targeted by airdrop hunters—malicious actors who exploit token distribution systems through address farming, automated scripts, and behavioral camouflage. While prior work such as ARTEMIS leverages multimodal features and local transaction patterns to detect such behavior, it lacks a global understanding of wallet influence in the transaction graph. In this paper, we propose an enhanced detection framework that augments the ARTEMIS by incorporating PageRank-based global centrality as an additional structural feature. This allows the model to better distinguish superficially active wallets from those with broader influence in the network. We evaluate our method on real-world Non-Fungible Token (NFT) data from the Blur marketplace and achieve state-of-the-art performance. Furthermore, a feature substitution experiment reveals that simple degree-based features alone can achieve near-perfect performance, even outperforming PageRank, suggesting that the labels are strongly coupled with topological properties. These findings highlight both the effectiveness of structural augmentation and the potential risks of shortcut learning in graph-based detection systems.

Lane-level high-definition map (HDMap) automatic construction from aerial imagery can significantly enhance the efficiency of HDMap production. However, some problems limit the application of aerial imagery in HDMap construction. First, the bird’s eye view of aerial imagery often causes occlusion, which leads to interrupted map element predictions. In addition, existing methods for aerial imagery-based HDMap construction can only construct road-level map and the postprocessing is complicated. To solve these problems, we propose AerialHDMapper, an end-to-end framework that predicts lane-level vectorized map elements directly requiring no postprocessing. The framework improves prediction accuracy in occluded areas through the introduction of the directional attention (DirAttn) module. AerialHDMapper is mainly composed of two parts: the image encoder and the polyline generator. The image encoder integrates DirAttn modules, which leverage information along and perpendicular to the direction of a map element, thereby improving the representation of the current position. The polyline generator employs a modified transformer architecture to identify long-sequence irregular map elements. Experimental results demonstrate that AerialHDMapper outperforms existing methods on both the CARLA Simulator Dataset and the Aerial Argoverse2 Dataset we collected. Extensive ablation experiments further reveal that our approach exhibits strong robustness and generalization capabilities.

Semantic segmentation of airborne laser scanning (ALS) point clouds is a valuable yet challenging task in remote sensing. When processing large-scale ALS scenes, it is necessary to partition them into smaller blocks for ease of handling. However, this partitioning introduces a challenge in capturing the ample spatial context within each block to adequately recognize the objects with a significant spatial span. This limitation becomes particularly pronounced when relying solely on the 3D representations as the input of nerual networks. To incorporate sufficient contextual information in ALS data semantic segmentation, we propose a multi-modal-based segmentation framework called projection-based context embedding (PCE) in this study. PCE effectively combines the advantages of 2D image and 3D point-voxel representations, which are the computational efficiency and the representation capability for fine-grained 3D geometries. The 2D projection is used to encode a large-scale semantic context, which is computationally expensive to be obtained using only pure 3D representation. Simultaneously, the sparse-point-voxel convolution (SPVConv) is employed to focus on learning 3D features from a small block of points centered on the large-scale context. Finally, to fully exploit the power of each modality, the embedding disentangling (ED) strategy is proposed additionally to combine the context embedding from the 2D image with 3D features for the final prediction. We demonstrate the state-of-the-art performance of PCE through extensive experiments on public large-scale ALS point cloud datasets.

Recent learning-based models excel in point cloud registration for low-overlap scenes but falter in scenarios with minimal overlap. In this article, we propose a novel method to address the extreme case of low-overlap registration: non-overlapping point cloud registration. This scenario involves input point clouds that do not have overlapping regions but are adjacent to each other after registration. While the practical application value of non-overlapping point cloud registration remains to be explored, we believe that researching this issue contributes to enhancing the performance of registration in scenarios with extremely low overlap. Abandoning conventional overlapping region detection, we directly generate the registered source point cloud with SCREAM, a generative adversarial network (GAN). The generator incorporates information from the target point cloud into the source point cloud’s features and generates the registered source point cloud. To further align the generated results with the target point cloud, we propose a differentiable renderer that renders both the target and predicted point clouds into depth maps. These depth maps are then used as inputs to a discriminator to determine whether the generated results align with the target point cloud. Rigid transformation can be directly estimated from the correspondences between the source and the generated point clouds, bypassing the need for detecting overlapping regions, feature matching, and RANSAC steps found in previous methods. Extensive experiments demonstrate that SCREAM not only outperforms common overlapping point cloud registration scenarios but also achieves a registration success rate of 52.6% for the first time in non-overlapping scenes. We also constructed a new indoor scene registration dataset, 3DZeroMatch, specifically designed to explore non-overlapping registration problems. Our code and the dataset 3DZeroMatch are accessible athttps://github.com/xujiabo/SCREAM/.

Automatic extraction of vector polygons of buildings from remotely sensed images is an important but difficult task. Recent existing methods based on deep learning usually adopt a multi-stage solution of semantic segmentation, contour detection, and polygon simplification. Such a long processing chain may lead to unreliable results as the boundary regularization and optimization processes are ultimately completed by utilizing low-level features, which ignores the potential of deep features in polygon generation. In this paper, we present an algorithm for directly extracting simplified polygons of buildings in remotely sensed images. The key of this task is the encoding of the polygon structure. PolyMapper [1] utilizes a recurrent neural network (RNN) to produce vertices of a polygon sequentially. Due to the limitation of RNN, this approach is unstable and difficult to deal with objects with complex shapes. In this work, we encode the polygon into a tensor representation and utilize a non-recurrent manner to recover the polygon structure. In our algorithm, two types of points are utilized, i.e., the corner point and the connecting point. Corner points are utilized to delineate the building outlines and form the vertices of the final polygon. Meanwhile, connecting points are sampled from the edges of the buildings for the assistance of the connection of the corner points. Furthermore, we predict the forward and backward directions of each keypoint in a polygon and propose a bidirectional tracing strategy for the polygon structure recovery. Our approach is simple, effective and robust. Experiments on public datasets demonstrate the superiority of the proposed algorithm. The code is made publicly available at https://github.com/sz94/bldvec.

Terrain scene clustering is a class of unsupervised methods for choosing suitable algorithms or parameters for airborne laser scanning (ALS) point cloud processing. Most existing point cloud clustering methods use hand-crafted features, such as viewpoint feature histogram (VFH), as the input of clustering algorithms. However, few studies on point cloud processing focused on terrain scene clustering via an unsupervised deep neural network. In the present study, we create a data set for terrain scene clustering in ALS point clouds. We also propose DPCC-Net, a deep point cloud clustering network via unsupervised deep learning that jointly learns the parameters of the network and the cluster task of extracted features. DPCC-Net iteratively groups the features extracted by the deep convolution neural network with the${k}$-means algorithm and uses the clustering result as the pseudo label to update the parameters of the network. We apply the proposed DPCC-Net to unsupervised training on a large terrain scene data set. The clustering result of DPCC-Net outperforms those of other typical methods.