- BB
Authors: Heipke, C.; Wang, X.; Advisor: -; Participants: - (2017) - Recently, low-cost 3D reconstruction based on images has become a popular focus of photogrammetry and computer vision research. Methods which can handle an arbitrary geometric setup of a large number of unordered and convergent images are of particular interest. However, determining the mutual overlap poses a considerable challenge. We propose a new method which was inspired by and improves upon methods employing random k-d forests for this task. Specifically, we first derive features from the images and then a random k-d forest is used to find the nearest neighbours in feature space. Subsequently, the degree of similarity between individual images, the image overlaps and thus images belonging to a common block are calculated as input to a structure-from-motion (sfm) pipeline. In ou...
|
- BB
Authors: Kimani, M.W.; Advisor: -; Participants: - (2017) - Accurate and consistent rainfall observations are vital for climatological studies in support of better agricultural and water management decision-making and planning. In East Africa, accurate rainfall estimation with an adequate spatial distribution is limited due to sparse rain gauge networks. Satellite rainfall products can potentially play a role in increasing the spatial coverage of rainfall estimates; however, their performance needs to be understood across space–time scales and factors relating to their errors. This study assesses the performance of seven satellite products: Tropical Applications of Meteorology using Satellite and ground-based observations (TAMSAT), African Rainfall Climatology And Time series (TARCAT), Climate Hazards Group InfraRed Precipitation with Statio...
|
- BB
Authors: Gehrung, J.; Advisor: -; Participants: - (2017) - Data recorded by mobile LiDAR systems (MLS) can be used for the generation and refinement of city models or for the automatic detection of long-term changes in the public road space. Since for this task only static structures are of interest, all mobile objects need to be removed. This work presents a straightforward but powerful approach to remove the subclass of moving objects. A probabilistic volumetric representation is utilized to separate MLS measurements recorded by a Velodyne HDL-64E into mobile objects and static background. The method was subjected to a quantitative and a qualitative examination using multiple datasets recorded by a mobile mapping platform. The results show that depending on the chosen octree resolution 87-95 % of the measurements are labeled correctly.
|
- BB
Authors: Tiede, D.; Advisor: -; Participants: - (2017) - Spatiotemporal analytics of multi-source Earth observation (EO) big data is a pre-condition for semantic content-based image retrieval (SCBIR). As a proof of concept, an innovative EO semantic querying (EO-SQ) subsystem was designed and prototypically implemented in series with an EO image understanding (EO-IU) subsystem. The EO-IU subsystem is automatically generating ESA Level 2 products (scene classification map, up to basic land cover units) from optical satellite data. The EO-SQ subsystem comprises a graphical user interface (GUI) and an array database embedded in a client server model. In the array database, all EO images are stored as a space-time data cube together with their Level 2 products generated by the EO-IU subsystem. The GUI allows users to (a) develop a conceptual ...
|
- BB
Authors: Drees, L.; Roscher, R.; Advisor: -; Participants: - (2017) - This paper focuses on the quantification of land cover fractions in an urban area of Berlin, Germany, using simulated hyperspectral EnMAP data with a spatial resolution of 30m×30m. For this, sparse representation is applied, where each pixel with unknown surface characteristics is expressed by a weighted linear combination of elementary spectra with known land cover class. The elementary spectra are determined from image reference data using simplex volume maximization, which is a fast heuristic technique for archetypal analysis. In the experiments, the estimation of class fractions based on the archetypal spectral library is compared to the estimation obtained by a manually designed spectral library by means of reconstruction error, mean absolute error of the fraction estimates, su...
|
- BB
Authors: -; Advisor: -; Participants: - (2017) - To support thedevelopment of EBVs by the Group on Earth Observations Biodiversity Observation Network (GEO BON), we identify11 key workflow steps that will operationalize the process of building EBV data products within and across researchinfrastructures worldwide. These workflow steps take multiple sequential activities into account, including identificationand aggregation of various raw data sources, data quality control, taxonomic name matching and statistical modellingof integrated data. We illustrate these steps with concrete examples from existing citizen science and professionalmonitoring projects, including eBird, the Tropical Ecology Assessment and Monitoring network, the Living PlanetIndex and the Baltic Sea zooplankton monitoring. The identified workflow steps are applica...
|
- BB
Authors: Paul, A.; Vogt, K.; Advisor: -; Participants: - (2017) - Supervised machine learning needs high quality, densely sampled and labelled training data. Transfer learning (TL) techniques have been devised to reduce this dependency by adapting classifiers trained on different, but related, (source) training data to new (target) data sets. A problem in TL is how to quantify the relatedness of a source quickly and robustly, because transferring knowledge from unrelated data can degrade the performance of a classifier. In this paper, we propose a method that can select a nearly optimal source from a large number of candidate sources. This operation depends only on the marginal probability distributions of the data, thus allowing the use of the often abundant unlabelled data. We extend this method to multi-source selection by optimizing a weighted...
|
- BB
Authors: Dewi, R.S.; Advisor: -; Participants: - (2017) - This paper addresses uncertainty modelling of shorelines by comparing fuzzy sets and random sets. Both methods quantify extensional uncertainty of shorelines extracted from remote sensing images. Two datasets were tested: pan-sharpened Pleiades with four bands (Pleiades) and pan-sharpened Pleiades stacked with elevation data as the fifth band (Pleiades + DTM). Both fuzzy sets and random sets model the spatial extent of shoreline including its uncertainty. Fuzzy sets represent shorelines as a margin determined by upper and lower thresholds and their uncertainty as confusion indices. They do not consider randomness. Random sets fit the mixed Gaussian model to the image histogram. It represents shorelines as a transition zone between water and non-water. Their extensional uncertainty i...
|
- BB
Authors: Beek, P.; Simon, B.; Advisor: -; Participants: - (2017) - We investigated submarine groundwater discharge (SGD), an important vector for many chemical elements that may impact the quality of the coastal environment. First, we acquired airborne thermal infrared (TIR) images to detect terrestrial groundwater inputs. Few studies have been conducted on SGD along the French Mediterranean coastline. The terrestrial groundwater spring inputs in La Palme and Salses-Leucate coastal lagoons are in the range (0.04–0.11) m3 s−1, ≤ 2% of the local river inputs. In comparison, total SGD estimates to La Palme lagoon (0.56–1.7 m3 s−1) suggest that the recirculation of lagoon water through the sediment is two orders of magnitude greater than the terrestrial groundwater inputs. This study demonstrates the application of airborne TIR remote sensing for detec...
|
- BB
Authors: Park, D.K.; Tegegne, G.; Advisor: -; Participants: - (2017) - We assessed daily streamflow predictions by applying two simple conceptual models and one complex model for four major gauged watersheds of the study area and compared these model’s capabilities in reproducing observed streamflow in the time and quantile domains. This study also addressed the sensitivity of the complex model for the number of partitioned subbasins. In the largest watershed, the performance of the complex model improved when the number of partitioned subbasins was increased. This finding indicates that the distributed models are especially applicable for the complex watershed because of its physical heterogeneity. In general, integrating these three models may be suitable for water resources assessment.
|
