Showing papers by "Norbert Pfeifer published in 2021"

A Comparison of Deep Learning Methods for Airborne Lidar Point Clouds Classification

Abstract: The success achieved by deep learning techniques in image labeling has triggered a growing interest in applying deep learning for three-dimensional point cloud classification. To provide better insights into different deep learning architectures and their applications to ALS point cloud classification, this article presents a comprehensive comparison among three state-of-the-art deep learning networks: PointNet++, SparseCNN, and KPConv, on two different ALS datasets. The performances of these three deep learning networks are compared w.r.t. classification accuracy, computation time, generalization ability as well as the sensitivity to the choices of hyper-parameters. Overall, we observed that PointNet++, SparseCNN, and KPConv all outperform Random Forest on the classification results. Moreover, SparseCNN leads to a slightly better classification result compared to PointNet++ and KPConv, while requiring less computation time and memory. At the same time, it shows a better ability to generalize and is less impacted by the different choices of hyper-parameters.

Potential of Sentinel-1 C-Band Time Series to Derive Structural Parameters of Temperate Deciduous Forests

Abstract: With the increasing occurrence of forest fires in the mid-latitudes and the alpine region, fire risk assessments become important in these regions. Fuel assessments involve the collection of information on forest structure as, e.g., the stand height or the stand density. The potential of airborne laser scanning (ALS) to provide accurate forest structure information has been demonstrated in several studies. Yet, flight acquisitions at the state level are carried out in intervals of typically five to ten years in Central Europe, which often makes the information outdated. The Sentinel-1 (S-1) synthetic aperture radar mission provides freely accessible earth observation (EO) data with short revisit times of 6 days. Forest structure information derived from this data source could, therefore, be used to update the respective ALS descriptors. In our study, we investigated the potential of S-1 time series to derive stand height and fractional cover, which is a measure of the stand density, over a temperate deciduous forest in Austria. A random forest (RF) model was used for this task, which was trained using ALS-derived forest structure parameters from 2018. The comparison of the estimated mean stand height from S-1 time series with the ALS derived stand height shows a root mean square error (RMSE) of 4.76 m and a bias of 0.09 m on a 100 m cell size, while fractional cover can be retrieved with an RMSE of 0.08 and a bias of 0.0. However, the predictions reveal a tendency to underestimate stand height and fractional cover for high-growing stands and dense areas, respectively. The stratified selection of the training set, which we investigated in order to achieve a more homogeneous distribution of the metrics for training, mitigates the underestimation tendency to some degree, yet, cannot fully eliminate it. We subsequently applied the trained model to S-1 time series of 2017 and 2019, respectively. The computed difference between the predictions suggests that large decreases in the forest height structure in this two-year interval become apparent from our RF-model, while inter-annual forest growth cannot be measured. The spatial patterns of the predicted forest height, however, are similar for both years (Pearson’s R = 0.89). Therefore, we consider that S-1 time series in combination with machine learning techniques can be applied for the derivation of forest structure information in an operational way.

Depth Measurement Bias in Pulsed Airborne Laser Hydrography Induced by Chromatic Dispersion

Abstract: In contrast to topographic laser scanning, laser hydrography must take into account the presence of two media. A pulsed laser beam, which enters the water from the air at an oblique angle, is refracted at the air–water boundary in the direction of the plumb line. This change in the direction described by Snellius’ law is caused by a slower speed of the light wave in the water, i.e., the phase velocity. Light scattering caused by turbidity gives rise to further deviations from the straight path. Together, the slower speed and the turbidity-induced path extension cause a longer pulse round trip time in the water than in the air. For an accurate measurement, it is important to correct this propagation time extension. It is a common practice to assume the phase velocity as the velocity for the laser pulses in water. In a dispersive medium, however, the phase velocity is only an approximation of the velocity of a pulse. In media with chromatic dispersion, the pulses propagate with a different velocity, i.e., the group velocity. In water, using the group velocity instead of the phase velocity reduces the range dependent bias of the depth measurement at a laser wavelength of 532 nm by more than 1.5%. We present an easy to perform an experiment, which shows that the group velocity differs so much from the phase velocity that this difference should be taken into account. We further discuss the use of group velocity to explain the depth bias using examples from the literature.

Multi-decadal (1953–2017) rock glacier morphodynamics analysed by high-resolution topographic data in the Upper Kauner Valley, Austria

Abstract: . Permafrost is being degraded worldwide due to the change in external forcing caused by climate change. This has also been shown to affect the morphodynamics of active rock glaciers. We studied these changes, depending on the analysis, on nine or eight active rock glaciers with different characteristics in multiple epochs between 1953 and 2017 in Kauner Valley, Austria. A combination of historical aerial photographs and airborne laser scanning data and their derivatives are used to analyse surface movement and 3D displacements. In general, the studied landforms show a significant acceleration of varying magnitude in the epoch 1997–2006 and a volume loss to varying degrees throughout the investigation period. Besides, we detect rock glaciers that show indication of inactivation. By analysing meteorological data (temperature, precipitation and snow cover onset and duration), we are able to identify possible links to these external forcing parameters. The combined investigation of horizontal and vertical 3D displacements shows that these are temporally decoupled on some rock glaciers. The catchment-wide survey further reveals that, despite the general trend, timing, magnitude and temporal peaks of morphodynamic changes indicate a slightly different sensitivity, response or response time of individual rock glaciers to fluctuations and changes in external forcing parameters.

Processing of nationwide topographic data for ensuring consistent river network representation

Abstract: Increasing river floods and infrastructure development in many parts of the world have created an urgent need for accurate high-resolution flood hazard mapping for more efficient flood risk management. Mapping accuracy hinges on the quality of the underlying Digital Terrain Model (DTM) and other spatial datasets. This article presents a processing strategy to ensure consistent adaption of countrywide spatial datasets to the requirements of hydraulic modelling. The suggested methods are automatized to a large extent and include (i) automatic fitting of river axis positions to the DTM, (ii) detection of culverts and obstacles in the river channel (iii) Smooth elimination of obstacles by interpolation along the river axes (iv) geometric detection of water-land borders and the top edge of embankments for (v) integration of the submerged river bed geometry into the DTM. The processing chain is applied to a river network (33880 km) and a DTM from Airborne Laser Scanning (ALS) with 1 m spatial resolution covering the entire territory of Austria ( ∼ 84000 km 2 ). Thus, countrywide consistency of data and methods is achieved along with high local relevance. Semi-automatic validation and extensive manual checks demonstrate that processing significantly improves the DTM with respect to topographic and hydraulic consistency. However, some open issues of automatic processing remain, e.g. in case of long underground river reaches.

Test Charts for Evaluating Imaging and Point Cloud Quality of Mobile Mapping Systems for Urban Street Space Acquisition

Abstract: Mobile mapping is in the process of becoming a routinely applied standard tool to support administration of cities. For ensuring the usability of the mobile mapping data it is necessary to have a practical method to evaluate the quality of different systems, which reaches beyond 3D accuracy of individual points. Such a method must be objective, easy to implement, and provide quantitative results to be used in tendering processes. We present such an approach which extracts quality figures for point density, point distribution, point cloud planarity, image resolution, and street sign legibility. In its practical application for the mobile mapping campaign of the City of Vienna (Austria) in 2020 the proposed test method proved to fulfill the above requirements. As an additional result, quality figures are reported for the panorama images and point clouds of three different mobile mapping systems.

A framework for generic spatial search in 3d point clouds

Abstract: . Modern data acquisition with active or passive photogrammetric imaging techniques generally results in 3D point clouds. Depending on the acquisition or processing method, the spacing of the individual points is either uniform or irregular. In the latter case, the neighbourhood definition like for digital images (4- or 8-neighbourhood, etc.) cannot be applied. Instead, analysis requires a local point neighbourhood. The local point neighbourhood with conventional k-nearest neighbour or fixed distance searches often produce sub-optimal results suffering from the inhomogeneous point distribution. In this article, we generalize the neighbourhood definition and present a generic spatial search framework which explicitly deals with arbitrary point patterns and aims at optimizing local point selection for specific processing tasks like interpolation, surface normal estimation and point feature extraction, spatial segmentation, and such like. The framework provides atomic 2D and 3D search strategies, (i) k-nearest neighbour, (ii) region query, (iii) cell based selection, and (iv) quadrant/octant based selection. It allows to freely combine the individual strategies to form complex, conditional search queries as well as specifically tailored point sub-selection. The benefits of such a comprehensive neighbourhood search approach are showcased for feature extraction and surface interpolation of irregularly distributed points.

Horizon Line Detection in Historical Terrestrial Images in Mountainous Terrain Based on the Region Covariance

Abstract: Horizon line detection is an important prerequisite for numerous tasks including the automatic estimation of the unknown camera parameters for images taken in mountainous terrain. In contrast to modern images, historical photographs contain no color information and have reduced image quality. In particular, missing color information in combination with high alpine terrain, partly covered with snow or glaciers, poses a challenge for automatic horizon detection. Therefore, a robust and accurate approach for horizon line detection in historical monochrome images in mountainous terrain was developed. For the detection of potential horizon pixels, an edge detector is learned based on the region covariance as texture descriptor. In combination with shortest path search the horizon in monochrome images is accurately detected. We evaluated our approach on 250 selected historical monochrome images in average dating back to 1950. In 85% of the images the horizon was detected with an error less than 10 pixels. In order to further evaluate the performance, an additional dataset consisting of modern color images was used. Our method, using only grayscale information, achieves comparable results with methods based on color information. In comparison with other methods using only grayscale information, accuracy of the detected horizons is significantly improved. Furthermore, the influence of color, choice of neighborhood for the shortest path calculation, and patch size for the calculation of the region covariance were investigated. The results show that both the availability of color information and increasing the patch size for the calculation of the region covariance improve the accuracy of the detected horizons.

Relationships between macro-fungal dark diversity and habitat parameters using LiDAR

Abstract: Despite the important role of fungi for ecosystems, relatively little is known about the factors underlying the dynamics of their diversity. Moreover, studies do not typically consider their dark diversity: the species absent from an otherwise suitable site. Here, we examined potential drivers of local fungal dark diversity in temperate woodland and open habitats using LiDAR and in-situ field measurements, combined with a systematically collected and geographically comprehensive macro-fungi and plant data set. For the first time, we also estimated species pools of fungi by considering both plant and fungi co-occurrences. The most important LiDAR variables for modelling fungal dark diversity were amplitude and echo ratio, which are both thought to represent vegetation structure. These results suggest that the local fungal dark diversity is highest in production forests like plantations and lowest in more open forests and in open habitats with little woody vegetation. Plant species richness was the strongest explanatory factor overall and negatively correlated with local fungal dark diversity. Soil fertility showed a positive relationship with dark diversity in open habitats. These findings may indicate that the local dark diversity of macro-fungi is highest in areas with a relatively high human impact (typically areas with low plant species richness and high soil fertility). Overall, this study brings novel insights into local macro-fungi dark diversity patterns, suggesting that a multitude of drivers related to both soil and vegetation act in concert to determine fungal dark diversity.