Very high resolution aerial image orthomosaics, point clouds, and elevation datasets of select permafrost landscapes in Alaska and northwestern Canada

Abstract. Fogo in the Cabo Verde archipelago off western Africa is one of the most prominent and active ocean island volcanoes on Earth, posing an important hazard both to local populations and at a regional level. The last eruption took place between 23 November 2014 and 8 February 2015 in the Chã das Caldeiras area at an elevation close to 1800 ma.s.l. The eruptive episode gave origin to extensive lava flows that almost fully destroyed the settlements of Bangaeira, Portela and Ilhéu de Losna. During December 2016 a survey of the Chã das Caldeiras area was conducted using a fixed-wing unmanned aerial vehicle (UAV) and real-time kinematic (RTK) global navigation satellite system (GNSS), with the objective of improving the terrain models and visible imagery derived from satellite platforms, from metric to decimetric resolution and accuracy. The main result is a very high resolution and quality 3D point cloud with a root mean square error of 0.08 m in X, 0.11 m in Y and 0.12 m in Z, which fully covers the most recent lava flows. The survey comprises an area of 23.9 km2 and used 2909 calibrated images with an average ground sampling distance of 7.2 cm. The dense point cloud, digital surface models and orthomosaics with 25 and 10 cm resolutions, a 50 cm spaced elevation contour shapefile, and a 3D texture mesh, as well as the full aerial survey dataset are provided. The delineation of the 2014/15 lava flows covers an area of 4.53 km2, which is smaller but more accurate than the previous estimates from 4.8 to 4.97 km2. The difference in the calculated area, when compared to previously reported values, is due to a more detailed mapping of the flow geometry and to the exclusion of the areas corresponding to kīpukas (outcrops surrounded by lava flows). Our study provides a very high resolution dataset of the areas affected by Fogo's latest eruption and is a case study supporting the advantageous use of UAV aerial photography surveys in disaster-prone areas. This dataset provides accurate baseline data for future eruptions, allowing for different applications in Earth system sciences, such as hydrology, ecology and spatial modelling, as well as to planning. The dataset is available for download at https://doi.org/10.5281/zenodo.4718520 (Vieira et al., 2021).

In the last decade, the combination of rapid development of low cost and small Unmanned Aerial Vehicles (UAVs), improved battery technology and conventional sensors (Optical and LiDAR) in terms of cost and dimensions, led to new opportunities in environmental remote-sensing and 3D surface modelling. A long term monitoring campaign was performed in Ricasoli village, in the Upper Arno river Valley (Tuscany, Italy), to understand the possibility of this rising technology to characterize and to monitor landslides. The RGB and multispectral imageries were analyzed and combined using SfM (Structure from Motion) software, in order to obtain high resolution orthomosaics, point clouds and 3D digital terrain models (DTM). The comparative analysis of the obtained DTMs allowed a very accurate reconstruction and mapping of the detected landslides. The collected data also allowed to precisely detect some slope portions prone to failure and to evaluate the area and volume of the involved masses as well as displacement rates.

Deep learning‐based single view 3D reconstruction is a hot topic in computer vision. However, predicting a more realistic 3D point cloud from a single image is an ill‐posed problem. In recent years, most of the 3D point cloud prediction methods based on single view are straight‐through structure, which will cause the loss of part of feature information and the loss of part of detail information of the resulting point clouds, which will lead to the unsatisfactory visual effect of reconstructed point clouds. In this paper, a Feature‐Enhanced 3D point clouds generation Network (3D‐FENet) from a single image is proposed. In order to enhance the feature information of RGB image, edge extraction module is adopted. In the process of point cloud generation, a point cloud pyramid is designed, which combines low resolution point cloud with high resolution point cloud to enhance the local details of the generated point clouds. In the fine‐tuning stage, the differential projection module is used to fine‐tune the whole network by 2D projection of reconstructed point clouds. Experimental results show that the performance of the authors’ proposed method is better than the state‐of‐the‐art studies.

<p>Photogrammetric surveys from UAV and LiDAR surveys are two techniques that allow for the production of very high resolution point clouds. The use of these techniques result in a detailed reconstruction of difficult-to-access environments such as underground cavities. A rigorous georeferencing of the acquired data allows for a comparison of the hypogean development of the cave to the overlying territory. This study presents a case of integration between these two techniques, applied to the risk assessment of the collapse of the vaults in a natural cavity in the Trieste Karst (north east Italy). This site is particularly delicate given that on the slope above the cave there is an abandoned stone quarry. In order to survey the quarry above the cave, a flight was performed with UAV, while the cave was surveyed with Laser Scan from the ground. The flight was made using a UAV DJI Phantom RTK, which carried a 20 Mpixel 1“ sensor camera. 8 ha of terrain was surveyed, capturing about 733 high resolution images and surveying 22 GCPs (Ground Control Point) with a GNSS RTK receiver. It was possible to reduce the number of GCPs, since the drone recorded the shooting positions very accurately with the on-board GPS RTK. Data were analyzed using Agisoft Metashape Professional to produce an orthophoto and a DSM (Digital Surface Model) with a ground resolution of 0.02 m and 0.04 m respectively. The point cloud has a density of 586 points/m<sup>2</sup>. The LiDaR survey was carried out using an ILRIS 3D ER laser scanner from Optec. The point cloud has a density of approximately 2500 points/m<sup>2</sup> and 5 stations were needed to cover the underground development of the cavity. The georeferencing of the data was carried out by roto-translation on geo-referenced benchmarks, surveyed with GPS RTK and total station. The point cloud was processed using Terrascan software (Terrasolid). The two point clouds were aligned, geo-referenced and combined using Polyworks software (Innovmetric), in order to check the thicknesses of the material present above the vault of the cave. The integration of epigean and hypogean data made it possible to identify some critical points related to a vault thickness of approximately 1.5 meters, located at the quarry square. This work made it possible to highlight critical issues difficult to detect without the integrated approach of these different survey methodologies.</p>

Point cloud is widely used because of its vectorized and compact information, but the unorder and discrete characteristics reduce the accuracy of point cloud reconstruction from images. In this paper, we propose a novel method to reconstruct high-resolution object point cloud by image redescription and point cloud upsampling. We first combine reconstruction and upsampling networks to generate high-resolution point cloud and achieve joint optimization through phased training. Then we present an image redescription mechanism to achieve the bidirectional correlation and enhance the semantic consistency between images and point clouds. The experiments on the ShapeNet dataset demonstrate the superiority of the proposed method over the state-of-the-art methods.

In this paper, we have described an accuracy analysis of MLS point clouds collected using the LARA3D prototype platform in an urban area. Accuracy of the MLS was achieved through comparison with other data sources more accurate that the studied system. The study has shown, that when compared with control points, collected by a Total Station, the prototype system LARA3D is able to produce data with an accuracy better then 0.3 m. However, taking into consideration the uncertainty in the identification of common points, this method is affected by man-made error and limited by point cloud resolutions. Meanwhile, the use of existing reference data, such as e.g. high resolution point clouds from static terrestrial laser scanning provides fast and reliable data evaluation. The subjective element of operator interpretation is also removed. Results achieved using ICP algorithm show, that our mobile mapping system suffers from limitations of the sensor quality and Kalman filter implementation. In the case of point clouds locally degraded, proper matching is impossible and the obtained result does not reflect the type and scale of deformation correctly. Meanwhile, another less time-consuming and more automated method for assessing data accuracy should be developed. That may be referred to using the existing spatial data as reference, such as e.g.: cadastre, ALS data, Topographic Data Base (TBD), Digital Terrain Model, orthophotos and so on.

In computer-vision-based human computer interaction (HCI), higher-quality signal leads to better system performance. In this paper, we develop a real-time high-resolution 3D object scanning system based on structured light illumination (SLI). Our system fuses depth information with RGB texture to reconstruct high-resolution 3D point cloud. The point cloud preserves accurate surface geometry of the object (e.g., finger postures of hands, facial expressions, etc). Respectively, for a 640 × 480 video stream, our system can generate phase and texture video at 1500 frames per second (fps) and produce full 3D point clouds at 300 fps. For gesture recognition, we propose to combine the module of robust face recognition with the module of 3D point cloud classification. Moreover, rather than extracting sophisticated features, we leverage the accurate reconstruction and classify each point cloud by directly matching the whole 3D surface geometry with the templates of different classes. The proposed recognition system is robust to the scaling, translation, rotation and texture of objects. Finally, utilizing the system, we contribute to the research community two large-scale high-resolution 3D point cloud databases, i.e., SLI 3D Hand Gesture Database and SLI 3D Face Database. The proposed point cloud recognition approach achieves recognition rates up to 98.0% over the gesture database and 88.2% over the face database in our pilot study.

Aerial imaging systems increasingly gain oblique viewing capabilities. Through these passive systems, photogrammetric 3D point clouds of a scene become available in addition to traditional vertical 2.5D information. In the field of urban reconstruction, this complementary information seeks for robust and automated fusion methods in order to derive 3D building geometry as well as topology in larger scales. It is sequentially shown how to get from façade planes over building footprints to roof reconstruction including overhangs. Façade planes are extracted from a photogrammetric high-resolution 3D point cloud. Local regression methods in 2D space are used to determine the local direction and a criterion for the local linearity of the point cloud. Based on these two parameters, the 3D point cloud is segmented according to which façade it belongs to. From the segmented point cloud, building footprints are extracted as polygons. Similar to cadaster information, those polygons, along with a traditional digital surface model (DSM), serve for one thing as the basis for overhang determination which is performed by fitting polynoms on the outside of façades and using their inflection points as overhang boundary. For another thing, they serve as roof areas which are segmented, topologically described and geometrically modelled. Again local regression methods are used but this time in 3D space in order to segment roof parts. Subsequently, the roof topology is derived using region growing methods. The final building models hold both, geometrical and topological properties.

Digital elevation models (DEM) were generated from oblique stereo-images acquired with a handheld digital camera. Two model scenarios are considered. Firstly, at local outcrop scale, with easy access, and distances between camera and outcrop varying between c. 40 m and c. 120 m, a very dense and high resolution point cloud was produced. The quality of the point cloud was evaluated against a terrestrial laser scan derived model of the same outcrop. The deviation between the two datasets varies between 0.02 m and 0.09. This is negligible for most geological purposes and illustrates the potential of using terrestrial photogrammetry at local outcrop scale as an alternative to lidar generated elevation data. Secondly, the method is explored at a regional scale, where a set of oblique stereo-images of a remotely located steep inaccessible mountain cliff was collected from a helicopter at a distance of c. 2–5 km under challenging and unfavourable conditions. The quality of the point cloud was evaluated against two elevation models extracted from conventional aerial photographs. Compared to a DEM extracted from monochrome aerial photographs, such as are often the only available topographic source for remote regions, a clear improvement in resolution is observed. Comparison with a DEM extracted from high resolution coloured aerial photographs shows the two digital elevation models to be very similar in resolution and with root mean square deviation (RMSE of 6.0 m).

Gunumuzde hizli gelisen uzay ve hava kaynakli uzaktan algilama teknolojileri, harita, jeoloji, cevre ve madencilik gibi araziye bagli muhendislik disiplinleri icin vazgecilmez hale gelmistir. Dijital fotogrametri ve hava kaynakli lazer tarama (Airborne Laser Scanning-ALS) ile cok yuksek cozunurluklu (Very High Resolution – VHR), hizli ulasilabilen hassas nokta bulutlari sayesinde topografik yuzey tanimlamasi kolay hale gelmistir. Fotogrametrik tekniklerin en guncellerinden biri olan optik donanimli insansiz hava araclari (IHA), farkli amaclar icin oldukca revactadir. IHA’lar dusuk ucus irtifasinin avantajini kullanarak yuksek cozunurluklu veri saglamaktadir. Bu calismada, Bulent Ecevit Universitesi Kampusu icerisinde yer alan bir insaat calismasi ve onun cevresel etkileri el yapimi bir optik IHA ile gozlemlenmistir. Uygulamada, topografyaya ait es yukseklik egrileri, sayisal arazi modeli (Digital Terrain Model - DTM) ve fark DTM’leri (Differential DTM- DiffDTM) kullanilarak zamansal degisimler belirlenmistir. DiffDTM’ler kullanilarak, topografya uzerindeki degisimler, es yukseklik egrilerinin morfolojik yapinin degisimini gosteren renk yukseklik skalasi ile gorsellestirilmistir.

We propose an octree guided neural network architecture and spherical convolutional kernel for machine learning from arbitrary 3D point clouds. The network architecture capitalizes on the sparse nature of irregular point clouds,and hierarchically coarsens the data representation with space partitioning. At the same time, the proposed spherical kernels systematically quantize point neighborhoods to identify local geometric structures in the data, while maintaining the properties of translation-invariance and asymmetry. We specify spherical kernels with the help of network neurons that in turn are associated with spatial locations.We exploit this association to avert dynamic kernel generation during network training that enables efficient learning with high resolution point clouds. The effectiveness of the proposed technique is established on the benchmark tasks of 3D object classification and segmentation, achieving competitive performance on ShapeNet and RueMonge2014 datasets.

In this paper, to solve the problem of lacking of road information caused by ground object occlusion, a registration and fusion method of high-resolution satellite image and vehicle point cloud data is proposed. Firstly, the road surface and crash barrier are extracted by using the filtering algorithm of joint gradient and elevation. Secondly, the Canny algorithm is used to extract road boundary on satellite images, and the plane is selected to extract linear points according to the elevation features of crash barrier and road boundary. Thirdly, the nearest neighbor point cloud iteration method is used to realize the matching of linear points with the same name. Finally, the high-resolution image and DEM are combined to generate a 3D model, and the vehicle point cloud data is registered with the three-dimensional model according to a rotation matrix, so as to improve the efficiency of high-precision map construction. The experimental results show that this method can effectively realize the registration of high resolution image and vehicle lidar data, combine the complementary advantages of high resolution image and point cloud data, improving the integrity of vehicle point cloud data, and alleviate the problem of high-precision map construction caused by occlusion to a certain extent.

An optimized fuzzy K-means clustering method for automated rock discontinuities extraction from point clouds

Abstract. Indoor mapping attracts more attention with the development of 2D and 3D camera and Lidar sensor. Lidar systems can provide a very high resolution and accurate point cloud. When aiming to reconstruct the static part of the scene, moving objects should be detected and removed which can prove challenging. This paper proposes a generic method to merge meshes produced from Lidar data that allows to tackle the issues of moving objects removal and static scene reconstruction at once. The method is adapted to a platform collecting point cloud from two Lidar sensors with different scan direction, which will result in different quality. Firstly, a mesh is efficiently produced from each sensor by exploiting its natural topology. Secondly, a visibility analysis is performed to handle occlusions (due to varying viewpoints) and remove moving objects. Then, a boolean optimization allows to select which triangles should be removed from each mesh. Finally, a stitching method is used to connect the selected mesh pieces. Our method is demonstrated on a Navvis M3 (2D laser ranger system) dataset and compared with Poisson and Delaunay based reconstruction methods.

With huge data acquisition progresses realized in the past decades and acquisition systems now able to produce high resolution grids and point clouds, the digitization of physical terrains becomes increasingly more precise. Such extreme quantities of generated and modeled data greatly impact computational performances on many levels of high-performance computing (HPC): storage media, memory requirements, transfer capability, and finally simulation interactivity, necessary to exploit this instance of big data. Efficient representations and storage are thus becoming “enabling technologies” in HPC experimental and simulation science. We propose HexaShrink, an original decomposition scheme for structured hexahedral volume meshes. The latter are used for instance in biomedical engineering, materials science, or geosciences. HexaShrink provides a comprehensive framework allowing efficient mesh visualization and storage. Its exactly reversible multiresolution decomposition yields a hierarchy of meshes of increasing levels of details, in terms of either geometry, continuous or categorical properties of cells. Starting with an overview of volume meshes compression techniques, our contribution blends coherently different multiresolution wavelet schemes in different dimensions. It results in a global framework preserving discontinuities (faults) across scales, implemented as a fully reversible upscaling at different resolutions. Experimental results are provided on meshes of varying size and complexity. They emphasize the consistency of the proposed representation, in terms of visualization, attribute downsampling and distribution at different resolutions. Finally, HexaShrink yields gains in storage space when combined to lossless compression techniques.