Search

Convolutional Neural Networks have been widely introduced to building rooftop segmentation using satellite and aerial imagery. Preparing efficient training data is still among the critical issues on this topic. Therefore, adopting available annotated cross-domain multisource dataset is needed. This paper evaluates the performance of fusing the state-of-art deep learning neural network architectures for cross-domain building rooftop segmentation. We have selected three semantic image segmentation neural networks, including Swin transformer, OCRNet and HRNet. The predictions from these three neural networks are combined with majority voting, max value and union fusion techniques, a refined building rooftop segmentation mask is therefore delivered. The experiments on two benchmark datasets show that the proposed fusion techniques outperform single models and other state-of-art cross-domain segmentation approaches.

The seagrass Posidonia oceanica is the main habitat-forming species of the coastal Mediterranean, providing millennial-scale ecosystem services including habitat provisioning, biodiversity maintenance, food security, coastal protection, and carbon sequestration. Meadows of this endemic seagrass species represent the largest carbon storage among seagrasses around the world, largely contributing to global blue carbon stocks. Yet, the slow growth of this temperate species and the extreme projected temperature and sea-level rise due to climate change increase the risk of reduction and loss of these services. Currently, there are knowledge gaps in its basin-wide spatially explicit extent and relevant accounting, therefore accurate and efficient mapping of its distribution and trajectories of change is needed. Here, we leveraged contemporary advances in Earth Observation—cloud computing, open satellite data, and machine learning—with field observations through a cloud-native geoprocessing framework to account the spatially explicit ecosystem extent of P. oceanica seagrass across its full bioregional scale. Employing 279,186 Sentinel-2 satellite images between 2015 and 2019, and a human-labeled training dataset of 62,928 pixels, we mapped 19,020 km2 of P. oceanica meadows up to 25 m of depth in 22 Mediterranean countries, across a total seabed area of 56,783 km2. Using 2,480 independent, field-based points, we observe an overall accuracy of 72%. We include and discuss global and region-specific seagrass blue carbon stocks using our bioregional seagrass extent estimate. As reference data collections, remote sensing technology and biophysical modelling improve and coalesce, such spatial ecosystem extent accounts could further support physical and monetary accounting of seagrass condition and ecosystem services, like blue carbon and coastal biodiversity. We envisage that effective policy uptake of these holistic seagrass accounts in national climate strategies and financing could accelerate transparent natural climate solutions and coastal resilience, far beyond the physical location of seagrass beds.

Highly accurate reference vehicle trajectories are required in the automotive domain e. g. for testing mobile GNSS devices. Common methods used to determine reference trajectories are based on the same working principles as the device under test and suffer from the same underlying error problems. In this paper, a new method to generate reference vehicle trajectories in real-world situations using simultaneously acquired aerial imagery from a helicopter is presented. This method requires independent height information which is coming from a LIDAR DTM and the relative height of the GNSS device. The reference trajectory is then derived by forward intersection of the vehicle position in each image with the DTM. In this context, the influence of all relevant error sources were analysed, like the error from the LIDAR DTM, from the sensor latency, from the semi-automatic matching of the vehicle marking, and from the image orientation. Results show that the presented method provides a tool for creating reference trajectories that is independent of the GNSS reception at the vehicle. Moreover, it can be demonstrated that the proposed method reaches an accuracy level of 10 cm, which is defined as necessary for certification and validation of automotive GNSS devices.

This study presents a novel method to identify optically deep water using purely spectral approaches. Optically deep waters, where the seabed is too deep for a bottom reflectance signal to be returned, is uninformative for seabed mapping. Furthermore, owing to the attenuation of light in the water column, submerged vegetation at deeper depths is easily confused with optically deep waters, thereby inducing misclassifications that reduce the accuracy of these seabed maps. While bathymetry data could mask out deeper areas, they are not always available or of sufficient spatial resolution for use. Without bathymetry data and based on the coastal aerosol blue green (1-2-3) bands of the Sentinel-2 imagery, this study investigates the use of band ratios and a false colour HSV transformation of both L1C and L2A images to separate optically deep and shallow waters across varying water quality over four tropical and temperate submerged sites: Tanzania, the Bahamas, the Caspian Sea (Kazakhstan) and the Wadden Sea (Denmark and Germany). Two supervised thresholds based on annotated reference data and an unsupervised Otsu threshold were applied. The band ratio group usually featured the best overall accuracies (OA), F1 scores and Matthews correlation coefficients, although the individual band combination might not perform consistently across different sites. Meanwhile, the saturation and hue band yielded close to best performance for the L1C and L2A images, featuring OA of up to 0.93 and 0.98, respectively, and a more consistent behaviour than the individual band ratios. Nonetheless, all these spectral methods are still susceptible to sunglint, the Sentinel-2 parallax effect, turbidity and water colour. Both supervised approaches performed similarly and were superior to the unsupervised Otsu’s method—the supervised methods featuring OA were usually over 0.70, while the unsupervised OA were usually under 0.80. In the absence of bathymetry data, this method could effectively remove optically deep water pixels in Sentinel-2 imagery and reduce the issue of dark pixel misclassification, thereby improving the benthic mapping of optically shallow waters and their seascapes.

Large-scale gridded population datasets available at the global or continental scale have become an important source of information in applications related to sustainable development. In recent years, the emergence of new population models has leveraged the inclusion of more accurate and spatially detailed proxy layers describing the built-up environment (e.g., built-area and building footprint datasets), enhancing the quality, accuracy and spatial resolution of existing products. However, due to the consistent lack of vertical and functional information on the built-up environ-ment, large-scale gridded population datasets that rely on existing built-up land proxies still report large errors of under-and overestimation, especially in areas with predominantly high-rise buildings or industrial/commercial areas, respectively. This research investigates, for the first time, the potential contributions of the new World Settlement Footprint—3D (WSF3D) dataset in the field of large-scale population modelling. First, we combined a Random Forest classifier with spatial metrics derived from the WSF3D to predict the industrial versus non-industrial use of settlement pixels at the Pan-European scale. We then examined the effects of including volume and settlement use information into frameworks of dasymetric population modelling. We found that the proposed classification method can predict industrial and non-industrial areas with overall accuracies and a kappa-coefficient of ~84% and 0.68, respectively. Additionally, we found that both, integrating volume and settlement use information considerably increased the accuracy of population estimates between 10% and 30% over commonly employed models (e.g., based on a binary settlement mask as input), mainly by eliminating systematic large overestimations in industrial/commercial areas. While the proposed method shows strong promise for overcoming some of the main limitations in large-scale population modelling, future research should focus on improving the quality of the WFS3D dataset and the classification method alike, to avoid the false detection of built-up settlements and to reduce misclassification errors of industrial and high-rise buildings.

Infrastructure and traffic monitoring are two of the most innovative applications for automatically extracting semantic information from aerial images which can be used for High-Definition (HD) mapping as a static map in autonomous driving. In this work, several novel Deep Learning algorithms have been developed for generating a detailed hybrid map as the combination of static and dynamic map over a large-area in a short time which includes direct lane-marking extraction from satellite imagery. Infrastruktur- und Verkehrsmonitoring sind zwei der innovativsten Anwendungen für die automatische Extraktion semantischer Informationen aus Luftbildern, die für High-Definition (HD) Mapping als statische Karte beim autonomen Fahren verwendet werden können. In dieser Arbeit wurden mehrere neuartige Deep-Learning-Algorithmen entwickelt, um in kurzer Zeit eine detaillierte hybride Karte als Kombination aus statischer und dynamischer Karte über ein großes Gebiet zu erstellen.

This dissertation deals with dense stereo matching of optical image data, in particular improving the accuracy of the well-established Semi-Global Matching (SGM) algorithm through different machine learning techniques. With a main focus on remote sensing data, three different algorithms are developed that improve the matching cost through self-supervised learning, optimize SGM regularization through classification, and provide efficient depth estimation through a pyramid-based end-to-end trainable deep learning network, which is particularly suitable for scenes with large height differences. Die Dissertation beschäftigt sich mit dichtem Stereomatching von optischen Bilddaten, insbesondere der Verbesserung der Genauigkeit des bewährten Semi-Global Matching (SGM) Algorithmus durch unterschiedliche Verfahren des maschinellen Lernens. Mit einem Hauptaugenmerk auf Fernerkundungsdaten werden drei unterschiedliche Algorithmen entwickelt, welche die Matching-Kosten durch selbstüberwachtes Lernen verbessern, die SGM-Regularisierung durch Klassifikation optimieren und eine effiziente Tiefenschätzung durch ein pyramidenbasiertes end-to-end trainierbares Deep Learning Netzwerk, welches sich insbesondere für Fernerkundungsdaten mit großen Höhenunterschieden eignet.