Article
Mathematics, Applied
Marco Caroccia
Summary: This work presents a compactness theorem for discrete functions on Poisson point clouds. It considers sequences with equibounded non-local p-Dirichlet energy, with the novelty being the computation of non-local energy in the intermediate-interaction regime.
NONLINEAR ANALYSIS-THEORY METHODS & APPLICATIONS
(2023)
Article
Plant Sciences
Wentao Liu, Chenglin Wang, De Yan, Weilin Chen, Lufeng Luo
Summary: This study explores a method for estimating grape feature parameters based on point cloud information: segment the grape point cloud using filtering and region growing algorithm, and register the complete grape point cloud model using an improved iterative closest point algorithm. The grape bunch surface was reconstructed using the Poisson algorithm after estimating model phenotypic size characteristics. Comparative analysis with existing methods shows that the proposed algorithm provides the closest estimation results to the measured parameters.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Environmental Sciences
Qi Liu, Shibiao Xu, Jun Xiao, Ying Wang
Summary: This paper introduces a novel sharp-feature-preserving reconstruction framework based on primitive detection, which accurately segments primitive patches, fits meshes in each patch, and splits overlapping meshes at the triangle level to ensure true sharpness and obtain lightweight mesh models. Experimental results show that our framework outperforms both the state-of-the-art learning-based primitive detection methods and traditional reconstruction methods. Moreover, our designed modules are plug-and-play, and can be combined with other point cloud processing tasks to achieve high-fidelity results.
Article
Computer Science, Artificial Intelligence
Hailiang Ye, Zijin Du, Feilong Cao
Summary: This paper proposes a novel method for obtaining 3D shape descriptors directly from point clouds, which utilizes local feature extraction and capsule network to avoid global pooling and effectively improve classification performance.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Qing Li, Cheng Wang, Chenglu Wen, Xin Li
Summary: This paper proposes DeepSIR, a novel learning-based iterative registration framework for real-world 3D LiDAR point clouds. The framework includes a front-end semantic feature extraction model, a point score that uses semantic and geometric information, an aggregation module to integrate the features and scores, and an iterative pipeline for exploring feature descriptions and optimizing poses. Experimental results demonstrate that DeepSIR achieves comparable performance to state-of-the-art methods and runs at a much faster speed. The source code will be made publicly available.
PATTERN RECOGNITION
(2023)
Article
Multidisciplinary Sciences
Zehao Zhou, Yichun Tai, Jianlin Chen, Zhijiang Zhang
Summary: This paper presents a local feature extraction network (LFE-Net) for point cloud analysis, which can learn geometric features suitable for various shape analysis problems. The network includes a local geometric relation module that learns high-dimensional local features to express the relation between points and their neighbors, and achieves state-of-the-art performances by combining features from multiple levels.
Article
Environmental Sciences
Tao Han, Gerardo Arturo Sanchez-Azofeifa
Summary: Lianas are increasingly dominating tropical forests due to climate change. Remote sensing techniques play a key role in separating lianas from their host trees, and this separation provides valuable insights into how tropical forests respond to climate and environmental change. In this study, a new machine learning method based on Random Forest and eXtreme Gradient Boosting algorithms is proposed to separate lianas and trees using Terrestrial Laser Scanning point clouds. The method is tested on tropical dry forest trees with different levels of liana infestation, and the results show high accuracy and recall rates. This method provides a flexible approach to extract lianas from 3D point clouds and facilitates new studies on the impact of lianas on tree and forest structures.
Article
Environmental Sciences
Tao Han, Gerardo Arturo Sanchez-Azofeifa
Summary: The study utilizes deep learning to separate leaf and woody components from terrestrial laser scanning data, comparing the performance of different neural network models. The results indicate that the multivariable time series (MTS) method has a higher accuracy than the univariable time series (UTS), while ResNet takes longer in model development.
Article
Humanities, Multidisciplinary
Lizhi Lou, Chaoxu Wei, Hangbin Wu, Chen Yang
Summary: This research proposes a method to extract and classify cave features from point cloud data of rockeries and improves the digitization quality of the rockeries. The experimental results show that this method can effectively extract and classify various types of caves, providing possibilities for future research.
Article
Computer Science, Artificial Intelligence
Saifullahi Aminu Bello, Cheng Wang, Naftaly Muriuki Wambugu, Jibril Muhammad Adam
Summary: This paper presents a new deep neural network FFPointNet that can exploit both local and global shape features of raw point clouds. By introducing a module called ChannelNet, the network achieved improved classification accuracy and intersection over union on popular datasets.
Article
Computer Science, Software Engineering
Shuaijun Chen, Jinxi Wang, Wei Pan, Shang Gao, Meili Wang, Xuequan Lu
Summary: This paper introduces a point cloud filtering method that considers both point distribution and feature preservation. The method incorporates a repulsion term and a data term in energy minimization to achieve this. The repulsion term is responsible for point distribution, while the data term aims to approximate noisy surfaces while preserving geometric features. The method is capable of handling models with fine-scale and sharp features, and experimental results demonstrate its ability to quickly yield good results with relatively uniform point distribution.
COMPUTATIONAL VISUAL MEDIA
(2023)
Article
Computer Science, Software Engineering
Zheng Liu, Xiaopeng Xin, Zheng Xu, Weijie Zhou, Chunxue Wang, Renjie Chen, Ying He
Summary: In this paper, a robust and reliable approach for geometric feature detection on surfaces within point clouds is presented. The approach accurately captures local surface variations at different feature sizes. By defining a bilateral weighted centroid projection-based metric, surface deviations are quantified. A structure-to-detail feature perception algorithm is proposed to accurately locate geometric features of varying sizes, and tensor analysis is used to extract boundary features. Experimental results demonstrate the effectiveness and versatility of the method in identifying a wide range of geometric characteristics within point clouds.
COMPUTER-AIDED DESIGN
(2023)
Article
Computer Science, Software Engineering
Dong Xiao, Zuoqiang Shi, Siyu Li, Bailin Deng, Bin Wang
Summary: This paper proposes a novel approach to orient point cloud normals by incorporating isovalue constraints to the Poisson equation, achieving consistent orientation for geometric algorithms based on point clouds. The method optimizes normals and implicit functions simultaneously to solve the global consistency problem. It demonstrates high performance and scalability, making it applicable to non-uniform and noisy data with varying sampling densities, artifacts, multiple connected components, and nested surfaces.
COMPUTER AIDED GEOMETRIC DESIGN
(2023)
Article
Computer Science, Artificial Intelligence
Xinhong Meng, Lei Zhu, Hailiang Ye, Feilong Cao
Summary: This paper proposes a multi-level interaction perception method for two-stage partial-to-partial point cloud registration, which hierarchically captures discriminative structural features by the interaction of local details and global features, and improves the perception of locality in the early information exchange. A spatial overlap-aware transformer is constructed to highlight the common regions of the point cloud while perceiving its global information, thus obtaining overlap constraints with high confidence between source and target point clouds.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2023)
Article
Computer Science, Software Engineering
Yechao Wang, Jinming Cao, Yangyan Li, Changhe Tu
Summary: An adaptive permutation module (APM) is proposed to address the order ambiguity in point cloud classification task, demonstrating its superiority through experiments and its flexibility to be integrated into other state-of-the-art approaches for performance improvement.
COMPUTERS & GRAPHICS-UK
(2021)
Article
Archaeology
Yuhe Zhang, Kang Li, Xiaoxue Chen, Shunli Zhang, Guohua Geng
JOURNAL OF CULTURAL HERITAGE
(2018)
Article
Computer Science, Information Systems
Shunli Zhang, Guohua Geng, Guohua Cao, Yuhe Zhang, Baodong Liu, Xu Dong
Article
Computer Science, Artificial Intelligence
Shunli Zhang, Xiangkui Zhang, Mingxiu Tuo, Haibo Zhang, Yuhe Zhang
Summary: This paper proposes a method for calculating the system matrix for Kohler's projection, which can achieve high accuracy in iterative reconstruction and significantly improve the quality of reconstruction.
JOURNAL OF AMBIENT INTELLIGENCE AND HUMANIZED COMPUTING
(2022)
Article
Computer Science, Information Systems
Yuhe Zhang, Xiaoning Liu, Chunhui Li, Jiabei Hu, Guohua Geng, Shunli Zhang
Article
Computer Science, Software Engineering
Rafael Maio, Tiago Araujo, Bernardo Marques, Andre Santos, Pedro Ramalho, Duarte Almeida, Paulo Dias, Beatriz Sousa Santos
Summary: Augmented Reality (AR) is a crucial technology in Industry 4.0 and smart manufacturing, particularly in the field of data monitoring. In this study, we developed a Pervasive AR tool for data monitoring, along with a web application for comparison purposes. User studies were conducted to gather feedback and evaluate the effectiveness of the systems, confirming the potential of Pervasive AR for data monitoring.
COMPUTERS & GRAPHICS-UK
(2024)
Article
Computer Science, Software Engineering
Berk Cebeci, Mehmet Bahadir Askin, Tolga K. Capin, Ufuk Celikcan
Summary: Despite advances in virtual reality technologies, extended VR sessions with head-mounted displays (HMDs) still face challenges in terms of comfort. In this study, a methodology using gaze-directed and visual saliency-guided paradigms for automatic stereo camera control in real-time interactive VR was proposed. The results showed that the gaze-directed approach outperformed the saliency-guided approach, both improving the overall depth feeling without hindering visual comfort in the tested virtual environments (VEs).
COMPUTERS & GRAPHICS-UK
(2024)
Article
Computer Science, Software Engineering
Ali Egemen Tasoren, Ufuk Celikcan
Summary: By developing the NOVAction engine, we have created the NOVAction23 dataset, which consists of highly diversified and photorealistic synthetic human action sequences. This dataset is significant in improving the performance of human action recognition.
COMPUTERS & GRAPHICS-UK
(2024)