- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2018 24th International Conference on Pattern Recognition (ICPR) - Beijing, China (2018.8.20-2018.8.24)] 2018 24th International Conference on Pattern Recognition (ICPR) - Accurate 3-D Reconstruction with RGB-D Cameras using Depth Map Fusion and Pose Refinement
摘要： Depth map fusion is an essential part in both stereo and RGB-D based 3-D reconstruction pipelines. Whether produced with a passive stereo reconstruction or using an active depth sensor, such as Microsoft Kinect, the depth maps have noise and may have poor initial registration. In this paper, we introduce a method which is capable of handling outliers, and especially, even significant registration errors. The proposed method first fuses a sequence of depth maps into a single non-redundant point cloud so that the redundant points are merged together by giving more weight to more certain measurements. Then, the original depth maps are re-registered to the fused point cloud to refine the original camera extrinsic parameters. The fusion is then performed again with the refined extrinsic parameters. This procedure is repeated until the result is satisfying or no significant changes happen between iterations. The method is robust to outliers and erroneous depth measurements as well as even significant depth map registration errors due to inaccurate initial camera poses.
关键词： point cloud,3-D reconstruction,RGB-D cameras,pose refinement,depth map fusion,registration errors
更新于2025-09-23 15:23:52
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Gaussian Decomposition of LiDAR Waveform Data Simulated by Dart
摘要： Light Detection And Ranging (LiDAR) techniques have been extensively applied in spaceborne, airborne and ground-based platforms. Understanding LiDAR data requires modeling approaches that can precisely account for the physical interactions between the emitted laser pulse and reflecting targets. Diverse LiDAR data types arise from different systems, platforms, and applications. However, most existing physical models consider only single pulse configurations to simulate large footprint LiDAR waveforms, which do not correspond to standard data formats. Hence, in many cases, model outputs are not well adapted to research conducted with actual LiDAR systems, especially for Aerial and Terrestrial Laser Scanning (ALS and TLS) systems. The Discrete Anisotropic Radiation Transfer (DART) model provides accurate and efficient simulations of multiple LiDAR pulses from all platform types. This paper presents the latest development of the DART LiDAR module: Gaussian decomposition of the simulated ALS and TLS waveforms followed by the provision of LiDAR point cloud and waveforms in text and standard ASPRS LAS formats.
关键词： point cloud,DART,waveform,LiDAR,ALS,Gaussian decomposition,TLS
更新于2025-09-23 15:23:52
-
Using multi-scale and hierarchical deep convolutional features for 3D semantic classification of TLS point clouds
摘要： Point cloud classification, which provides meaningful semantic labels to the points in a point cloud, is essential for generating three-dimensional (3D) models. Its automation, however, remains challenging due to varying point densities and irregular point distributions. Adapting existing deep-learning approaches for two-dimensional (2D) image classification to point cloud classification is inefficient and results in the loss of information valuable for point cloud classification. In this article, a new approach that classifies point cloud directly in 3D is proposed. The approach uses multi-scale features generated by deep learning. It comprises three steps: (1) extract single-scale deep features using 3D convolutional neural network (CNN); (2) subsample the input point cloud at multiple scales, with the point cloud at each scale being an input to the 3D CNN, and combine deep features at multiple scales to form multi-scale and hierarchical features; and (3) retrieve the probabilities that each point belongs to the intended semantic category using a softmax regression classifier. The proposed approach was tested against two publicly available point cloud datasets to demonstrate its performance and compared to the results produced by other existing approaches. The experiment results achieved 96.89% overall accuracy on the Oakland dataset and 91.89% overall accuracy on the Europe dataset, which are the highest among the considered methods.
关键词： point cloud,multi-scale,classification,3D,Deep learning
更新于2025-09-23 15:23:52
-
Derivation of space-resolved normal joint spacing and in situ block size distribution data from terrestrial LIDAR point clouds in a rugged Alpine relief (Kühtai, Austria)
摘要： Terrestrial laserscan (TLS) surveys allow the geological investigation of rock slopes, which cannot be measured by direct surveys because of inaccessibility, high hazard potential or excessive effort. The normal joint spacing and the in situ block size distribution are relevant properties for rock mass characterisation but are commonly evaluated statistically or at small regions only. This study presents the jointing characterisation of an Alpine rock slope by both scanline data and a new, automated analysis of point cloud data. The slope, located in the L?ngental (Austria), is characterised by a rugged Alpine relief and granodioritic gneisses fractured by non-persistent joints. The scanline data and the TLS surveys were used to investigate joint set orientations, normal joint spacings and in situ block sizes. Area-wide maps of rock slope properties were prepared from the results of the point cloud analysis. The general results derived from the point clouds are in good agreement with the scanline data. The space-resolved maps show larger block sizes in some of the higher ranging sub-regions and small block sizes in tectonically formed gullies, as well as various local variations. These visualisations are much more beneficial for most rock mechanical questions than common statistical data evaluation approaches using pre-defined sub-regions, which are treated as homogenous areas and thus are missing space-resolved information.
关键词： Point cloud analysis,Terrestrial laserscan,Normal joint spacing,Austria,In situ block size distribution,Joint characterisation
更新于2025-09-23 15:23:52
-
Cloud removal in remote sensing images using nonnegative matrix factorization and error correction
摘要： In the imaging process of optical remote sensing platforms, clouds are an inevitable barrier to the effective observation of sensors. To recover the original information covered by the clouds and the accompanying shadows, a nonnegative matrix factorization (NMF) and error correction method (S-NMF-EC) is proposed in this paper. Firstly, a cloud-free fused reference image is obtained by a reference image and two or more low-resolution images using the spatial and temporal nonlocal filter-based data fusion model (STNLFFM). Secondly, the cloud-free fused reference image is used to remove the cloud cover of the cloud-contaminated image based on NMF. Finally, the cloud removal result is further improved by error correction. It is worth noting that cloud detection is not required by S-NMF-EC, and the cloud-free information of the cloud-contaminated image is maximally retained. Both simulated and real-data experiments were conducted to validate the proposed S-NMF-EC method. Compared with other cloud removal methods, the results demonstrate that S-NMF-EC is visually and quantitatively effective (correlation coefficients ≥ 0.99) for the removal of thick clouds, thin clouds, and shadows.
关键词： Nonnegative matrix factorization,Multitemporal,Optical remote sensing image,Error correction,Cloud removal
更新于2025-09-23 15:23:52
-
[IEEE 2018 26th International Conference on Geoinformatics - Kunming (2018.6.28-2018.6.30)] 2018 26th International Conference on Geoinformatics - Multilevel Solar Potential Analysis of Building Based on Ubiquitous Point Clouds
摘要： Solar potential analysis is essential for Building Information Modeling (BIM) applications, like photovoltaic installation. The estimation of solar potential on rooftops has been widely discussed, whereas the study on fa?ade is still limited. Benefit from the development of various sensors, ubiquitous point clouds are now widely and easily captured by photogrammetry, laser scanning or other technologies, to represent the building geometry. This paper proposes a method for solar potential analysis on both rooftops and fa?ades, using ubiquitous point cloud collected by Unmanned Aerial Vehicle (UAV) and Terrestrial Laser Scanning (TLS). One building with different orientations is selected for the case study. Results show that the proposed method is valid for multilevel solar potential analysis of buildings.
关键词： laser scanning,solar potential,ubiquitous point cloud,Building Information Modeling (BIM)
更新于2025-09-23 15:23:52
-
Cloud detection and classification based on MAX-DOAS observations
摘要： Multi-axis differential optical absorption spectroscopy (MAX-DOAS) observations of aerosols and trace gases can be strongly influenced by clouds. Thus, it is important to identify clouds and characterise their properties. In this study we investigate the effects of clouds on several quantities which can be derived from MAX-DOAS observations, like radiance, the colour index (radiance ratio at two selected wavelengths), the absorption of the oxygen dimer O4 and the fraction of inelastically scattered light (Ring effect). To identify clouds, these quantities can be either compared to their corresponding clear-sky reference values, or their dependencies on time or viewing direction can be analysed. From the investigation of the temporal variability the influence of clouds can be identified even for individual measurements. Based on our investigations we developed a cloud classification scheme, which can be applied in a flexible way to MAX-DOAS or zenith DOAS observations: in its simplest version, zenith observations of the colour index are used to identify the presence of clouds (or high aerosol load). In more sophisticated versions, other quantities and viewing directions are also considered, which allows subclassifications like, e.g., thin or thick clouds, or fog. We applied our cloud classification scheme to MAX-DOAS observations during the Cabauw intercomparison campaign of Nitrogen Dioxide measuring instruments (CINDI) campaign in the Netherlands in summer 2009 and found very good agreement with sky images taken from the ground and backscatter profiles from a lidar.
关键词： cloud detection,CINDI campaign,radiative transfer,cloud classification,MAX-DOAS
更新于2025-09-23 15:22:29
-
[IEEE 2018 20th International Conference on Transparent Optical Networks (ICTON) - Bucharest (2018.7.1-2018.7.5)] 2018 20th International Conference on Transparent Optical Networks (ICTON) - Supporting Low-Latency Applications through Hybrid Cost-Optimised Cloudlet Placement
摘要： Low-latency applications such as augmented reality, cognitive assistance, and context-aware computation are better supported by augmenting cloud networks with cloudlet enabled edge computing solution. This overcomes the high-transmission latency arising from the edge devices in the access segment to cloud servers located in the core. In this talk, we will review our recently proposed hybrid cost-optimization framework for optimal cloudlet placement over existing passive optical access networks, subject to capacity and latency constraints. In our work, we formulate a mixed-integer non-linear program to identify ideal locations (either at the central office (CO), remote node (RN), or in the field) for cloudlet placement over three deployment areas of differing population densities. Our results point to the fact that the installation of more RN and CO-located cloudlets will yield an improved cost optimal solution than the installation of field cloudlets alone, and that the percentage of the incremental energy budget arising from the installation of cloudlets are low.
关键词： non-linear mixed-integer programming,passive optical access network,cloud computing,low-latency,Tactile Internet,cloudlets,cost-optimization
更新于2025-09-23 15:22:29
-
[IEEE 2018 25th International Conference on Mechatronics and Machine Vision in Practice (M2VIP) - Stuttgart, Germany (2018.11.20-2018.11.22)] 2018 25th International Conference on Mechatronics and Machine Vision in Practice (M2VIP) - 3D Point Cloud Coarse Registration based on Convex Hull Refined by ICP and NDT
摘要： Non-rigid registration is a crucial step for many applications such as motion tracking, model retrieval, and object recognition. The accuracy of these applications is highly dependent on the initial position used in registration step. In this paper we propose a novel Convex Hull Aided Coarse Registration refined by two algorithms applied on projected points.Firstly,the proposed approach uses a statistical method to find the best plane that represents each point cloud. Secondly, all the points of each cloud are projected onto the corresponding planes. Then, two convex hulls are extracted from the two projected point sets and then matched optimally. Next, the non-rigid transformation from the reference to the model is robustly estimated through minimizing the distance between the matched point's pairs of the two convex hulls.Finally, this transformation estimation is refined by two methods. The first one is the refinement of coarse registration by Iterative Closest Point (ICP). The second one consists of the refinement of coarse registration by the Normal Distribution Transform (NDT). An experimental study ,carried out on several clouds, shows that the refinement of coarse registration with ICP gives, in the most cases, a better result than refinement with NDT.
关键词： Iterative Closest Point (ICP),Convex Hull,Normal Distribution Transform (NDT),Non rigid registration,3D point cloud,Principal Component Analysis (PCA)
更新于2025-09-23 15:22:29
-
[IEEE 2018 IEEE International Conference on Intelligent Transportation Systems (ITSC) - Maui, HI, USA (2018.11.4-2018.11.7)] 2018 21st International Conference on Intelligent Transportation Systems (ITSC) - Vehicle Detection and Localization using 3D LIDAR Point Cloud and Image Semantic Segmentation
摘要： This paper presents a real-time approach to detect and localize surrounding vehicles in urban driving scenes. We propose a multimodal fusion framework that processes both 3D LIDAR point cloud and RGB image to obtain robust vehicle position and size in a Bird's Eye View (BEV). Semantic segmentation from RGB images is obtained using our efficient Convolutional Neural Network (CNN) architecture called ERFNet. Our proposal takes advantage of accurate depth information provided by LIDAR and detailed semantic information processed from a camera. The method has been tested using the KITTI object detection benchmark. Experiments show that our approach outperforms or is on par with other state-of-the-art proposals but our CNN was trained in another dataset, showing a good generalization capability to any domain, a key point for autonomous driving.
关键词： localization,ERFNet,image semantic segmentation,KITTI,autonomous driving,vehicle detection,CNN,point cloud,multimodal fusion,3D LIDAR
更新于2025-09-23 15:22:29