@article{VermaKummerowDenker2018,
  author    = {Verma, Meetu and Kummerow, P. and Denker, Carsten},
  title     = {On the extent of the moat flow in axisymmetric sunspots},
  series = {Astronomische Nachrichten = Astronomical notes},
  volume    = {339},
  journal   = {Astronomische Nachrichten = Astronomical notes},
  number    = {4},
  publisher = {Wiley-VCH},
  address   = {Weinheim},
  issn      = {0004-6337},
  doi       = {10.1002/asna.201813482},
  pages     = {268 -- 276},
  year      = {2018},
  abstract  = {Unipolar, axisymmetric sunspots are figuratively called "theoretician's sunspots" because their simplicity supposedly makes them more suitable for theoretical descriptions or numerical models. On November 18, 2013, a very large specimen (active region NOAA 11899) crossed the central meridian of the sun. The moat flow associated with this very large spot is quantitatively compared to that of a medium and a small sunspot to determine the extent of the moat flow in different environments. We employ continuum images and magnetograms of the Helioseismic and Magnetic Imager (HMI) as well as extreme ultraviolet (EUV) images at λ160 nm of the Atmospheric Imaging Assembly (AIA), both on board the Solar Dynamics Observatory (SDO), to measure horizontal proper motions with Local Correlation Tracking (LCT) and flux transport velocities with the Differential Affine Velocity Estimator (DAVE). We compute time-averaged flow maps (±6 hr around meridian passage) and radial averages of photometric, magnetic, and flow properties. Flow fields of a small- and a medium-sized axisymmetric sunspot provide the context for interpreting the results. All sunspots show outward moat flow and the advection of moving magnetic features (MMFs). However, the extent of the moat flow varies from spot to spot, and a correlation of flow properties with size is tenuous, if at all present. The moat flow is asymmetric and predominantly in the east-west direction, whereby deviations are related to the tilt angle of the sunspot group as well as to the topology and activity level of the trailing plage.},
  language  = {en}
}
@misc{SoechtingTrapp2020,
  author    = {S{\"o}chting, Maximilian and Trapp, Matthias},
  title     = {Controlling image-stylization techniques using eye tracking},
  series = {Postprints der Universit{\"a}t Potsdam : Reihe der Digital Engineering Fakult{\"a}t},
  journal   = {Postprints der Universit{\"a}t Potsdam : Reihe der Digital Engineering Fakult{\"a}t},
  number    = {7},
  doi       = {10.25932/publishup-52471},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-524717},
  pages     = {12},
  year      = {2020},
  abstract  = {With the spread of smart phones capable of taking high-resolution photos and the development of high-speed mobile data infrastructure, digital visual media is becoming one of the most important forms of modern communication. With this development, however, also comes a devaluation of images as a media form with the focus becoming the frequency at which visual content is generated instead of the quality of the content. In this work, an interactive system using image-abstraction techniques and an eye tracking sensor is presented, which allows users to experience diverting and dynamic artworks that react to their eye movement. The underlying modular architecture enables a variety of different interaction techniques that share common design principles, making the interface as intuitive as possible. The resulting experience allows users to experience a game-like interaction in which they aim for a reward, the artwork, while being held under constraints, e.g., not blinking. The co nscious eye movements that are required by some interaction techniques hint an interesting, possible future extension for this work into the field of relaxation exercises and concentration training.},
  language  = {en}
}
@phdthesis{Shekhar2023,
  author    = {Shekhar, Sumit},
  title     = {Image and video processing based on intrinsic attributes},
  doi       = {10.25932/publishup-62004},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-620049},
  school      = {Universit{\"a}t Potsdam},
  pages     = {xii, 143},
  year      = {2023},
  abstract  = {Advancements in computer vision techniques driven by machine learning have facilitated robust and efficient estimation of attributes such as depth, optical flow, albedo, and shading. To encapsulate all such underlying properties associated with images and videos, we evolve the concept of intrinsic images towards intrinsic attributes. Further, rapid hardware growth in the form of high-quality smartphone cameras, readily available depth sensors, mobile GPUs, or dedicated neural processing units have made image and video processing pervasive. In this thesis, we explore the synergies between the above two advancements and propose novel image and video processing techniques and systems based on them. To begin with, we investigate intrinsic image decomposition approaches and analyze how they can be implemented on mobile devices. We propose an approach that considers not only diffuse reflection but also specular reflection; it allows us to decompose an image into specularity, albedo, and shading on a resource constrained system (e.g., smartphones or tablets) using the depth data provided by the built-in depth sensors. In addition, we explore how on-device depth data can further be used to add an immersive dimension to 2D photos, e.g., showcasing parallax effects via 3D photography. In this regard, we develop a novel system for interactive 3D photo generation and stylization on mobile devices. Further, we investigate how adaptive manipulation of baseline-albedo (i.e., chromaticity) can be used for efficient visual enhancement under low-lighting conditions. The proposed technique allows for interactive editing of enhancement settings while achieving improved quality and performance. We analyze the inherent optical flow and temporal noise as intrinsic properties of a video. We further propose two new techniques for applying the above intrinsic attributes for the purpose of consistent video filtering. To this end, we investigate how to remove temporal inconsistencies perceived as flickering artifacts. One of the techniques does not require costly optical flow estimation, while both provide interactive consistency control. Using intrinsic attributes for image and video processing enables new solutions for mobile devices - a pervasive visual computing device - and will facilitate novel applications for Augmented Reality (AR), 3D photography, and video stylization. The proposed low-light enhancement techniques can also improve the accuracy of high-level computer vision tasks (e.g., face detection) under low-light conditions. Finally, our approach for consistent video filtering can extend a wide range of image-based processing for videos.},
  language  = {en}
}
@phdthesis{Semmo2016,
  author    = {Semmo, Amir},
  title     = {Design and implementation of non-photorealistic rendering techniques for 3D geospatial data},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-99525},
  school      = {Universit{\"a}t Potsdam},
  pages     = {XVI, 155},
  year      = {2016},
  abstract  = {Geospatial data has become a natural part of a growing number of information systems and services in the economy, society, and people's personal lives. In particular, virtual 3D city and landscape models constitute valuable information sources within a wide variety of applications such as urban planning, navigation, tourist information, and disaster management. Today, these models are often visualized in detail to provide realistic imagery. However, a photorealistic rendering does not automatically lead to high image quality, with respect to an effective information transfer, which requires important or prioritized information to be interactively highlighted in a context-dependent manner. Approaches in non-photorealistic renderings particularly consider a user's task and camera perspective when attempting optimal expression, recognition, and communication of important or prioritized information. However, the design and implementation of non-photorealistic rendering techniques for 3D geospatial data pose a number of challenges, especially when inherently complex geometry, appearance, and thematic data must be processed interactively. Hence, a promising technical foundation is established by the programmable and parallel computing architecture of graphics processing units. This thesis proposes non-photorealistic rendering techniques that enable both the computation and selection of the abstraction level of 3D geospatial model contents according to user interaction and dynamically changing thematic information. To achieve this goal, the techniques integrate with hardware-accelerated rendering pipelines using shader technologies of graphics processing units for real-time image synthesis. The techniques employ principles of artistic rendering, cartographic generalization, and 3D semiotics—unlike photorealistic rendering—to synthesize illustrative renditions of geospatial feature type entities such as water surfaces, buildings, and infrastructure networks. In addition, this thesis contributes a generic system that enables to integrate different graphic styles—photorealistic and non-photorealistic—and provide their seamless transition according to user tasks, camera view, and image resolution. Evaluations of the proposed techniques have demonstrated their significance to the field of geospatial information visualization including topics such as spatial perception, cognition, and mapping. In addition, the applications in illustrative and focus+context visualization have reflected their potential impact on optimizing the information transfer regarding factors such as cognitive load, integration of non-realistic information, visualization of uncertainty, and visualization on small displays.},
  language  = {en}
}
@phdthesis{Muehlbauer2011,
  author    = {M{\"u}hlbauer, Felix},
  title     = {Entwurf, Methoden und Werkzeuge f{\"u}r komplexe Bildverarbeitungssysteme auf Rekonfigurierbaren System-on-Chip-Architekturen},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus-59923},
  school      = {Universit{\"a}t Potsdam},
  year      = {2011},
  abstract  = {Bildverarbeitungsanwendungen stellen besondere Anspr{\"u}che an das ausf{\"u}hrende Rechensystem. Einerseits ist eine hohe Rechenleistung erforderlich. Andererseits ist eine hohe Flexibilit{\"a}t von Vorteil, da die Entwicklung tendentiell ein experimenteller und interaktiver Prozess ist. F{\"u}r neue Anwendungen tendieren Entwickler dazu, eine Rechenarchitektur zu w{\"a}hlen, die sie gut kennen, anstatt eine Architektur einzusetzen, die am besten zur Anwendung passt. Bildverarbeitungsalgorithmen sind inh{\"a}rent parallel, doch herk{\"o}mmliche bildverarbeitende eingebettete Systeme basieren meist auf sequentiell arbeitenden Prozessoren. Im Gegensatz zu dieser "Unstimmigkeit" k{\"o}nnen hocheffiziente Systeme aus einer gezielten Synergie aus Software- und Hardwarekomponenten aufgebaut werden. Die Konstruktion solcher System ist jedoch komplex und viele L{\"o}sungen, wie zum Beispiel grobgranulare Architekturen oder anwendungsspezifische Programmiersprachen, sind oft zu akademisch f{\"u}r einen Einsatz in der Wirtschaft. Die vorliegende Arbeit soll ein Beitrag dazu leisten, die Komplexit{\"a}t von Hardware-Software-Systemen zu reduzieren und damit die Entwicklung hochperformanter on-Chip-Systeme im Bereich Bildverarbeitung zu vereinfachen und wirtschaftlicher zu machen. Dabei wurde Wert darauf gelegt, den Aufwand f{\"u}r Einarbeitung, Entwicklung als auch Erweiterungen gering zu halten. Es wurde ein Entwurfsfluss konzipiert und umgesetzt, welcher es dem Softwareentwickler erm{\"o}glicht, Berechnungen durch Hardwarekomponenten zu beschleunigen und das zu Grunde liegende eingebettete System komplett zu prototypisieren. Hierbei werden komplexe Bildverarbeitungsanwendungen betrachtet, welche ein Betriebssystem erfordern, wie zum Beispiel verteilte Kamerasensornetzwerke. Die eingesetzte Software basiert auf Linux und der Bildverarbeitungsbibliothek OpenCV. Die Verteilung der Berechnungen auf Software- und Hardwarekomponenten und die daraus resultierende Ablaufplanung und Generierung der Rechenarchitektur erfolgt automatisch. Mittels einer auf der Antwortmengenprogrammierung basierten Entwurfsraumexploration ergeben sich Vorteile bei der Modellierung und Erweiterung. Die Systemsoftware wird mit OpenEmbedded/Bitbake synthetisiert und die erzeugten on-Chip-Architekturen auf FPGAs realisiert.},
  language  = {de}
}
@article{KreowskyStabernack2021,
  author    = {Kreowsky, Philipp and Stabernack, Christian Benno},
  title     = {A full-featured FPGA-based pipelined architecture for SIFT extraction},
  series = {IEEE access : practical research, open solutions / Institute of Electrical and Electronics Engineers},
  volume    = {9},
  journal   = {IEEE access : practical research, open solutions / Institute of Electrical and Electronics Engineers},
  publisher = {Inst. of Electr. and Electronics Engineers},
  address   = {New York, NY},
  issn      = {2169-3536},
  doi       = {10.1109/ACCESS.2021.3104387},
  pages     = {128564 -- 128573},
  year      = {2021},
  abstract  = {Image feature detection is a key task in computer vision. Scale Invariant Feature Transform (SIFT) is a prevalent and well known algorithm for robust feature detection. However, it is computationally demanding and software implementations are not applicable for real-time performance. In this paper, a versatile and pipelined hardware implementation is proposed, that is capable of computing keypoints and rotation invariant descriptors on-chip. All computations are performed in single precision floating-point format which makes it possible to implement the original algorithm with little alteration. Various rotation resolutions and filter kernel sizes are supported for images of any resolution up to ultra-high definition. For full high definition images, 84 fps can be processed. Ultra high definition images can be processed at 21 fps.},
  language  = {en}
}
@article{KonakWegnerArnrich2020,
  author    = {Konak, Orhan and Wegner, Pit and Arnrich, Bert},
  title     = {IMU-Based Movement Trajectory Heatmaps for Human Activity Recognition},
  series = {Sensors},
  volume    = {20},
  journal   = {Sensors},
  number    = {24},
  publisher = {MDPI},
  address   = {Basel},
  issn      = {1424-8220},
  doi       = {10.3390/s20247179},
  pages     = {15},
  year      = {2020},
  abstract  = {Recent trends in ubiquitous computing have led to a proliferation of studies that focus on human activity recognition (HAR) utilizing inertial sensor data that consist of acceleration, orientation and angular velocity. However, the performances of such approaches are limited by the amount of annotated training data, especially in fields where annotating data is highly time-consuming and requires specialized professionals, such as in healthcare. In image classification, this limitation has been mitigated by powerful oversampling techniques such as data augmentation. Using this technique, this work evaluates to what extent transforming inertial sensor data into movement trajectories and into 2D heatmap images can be advantageous for HAR when data are scarce. A convolutional long short-term memory (ConvLSTM) network that incorporates spatiotemporal correlations was used to classify the heatmap images. Evaluation was carried out on Deep Inertial Poser (DIP), a known dataset composed of inertial sensor data. The results obtained suggest that for datasets with large numbers of subjects, using state-of-the-art methods remains the best alternative. However, a performance advantage was achieved for small datasets, which is usually the case in healthcare. Moreover, movement trajectories provide a visual representation of human activities, which can help researchers to better interpret and analyze motion patterns.},
  language  = {en}
}
@misc{KonakWegnerArnrich2021,
  author    = {Konak, Orhan and Wegner, Pit and Arnrich, Bert},
  title     = {IMU-Based Movement Trajectory Heatmaps for Human Activity Recognition},
  series = {Postprints der Universit{\"a}t Potsdam : Reihe der Digital Engineering Fakult{\"a}t},
  journal   = {Postprints der Universit{\"a}t Potsdam : Reihe der Digital Engineering Fakult{\"a}t},
  number    = {4},
  doi       = {10.25932/publishup-48779},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-487799},
  pages     = {17},
  year      = {2021},
  abstract  = {Recent trends in ubiquitous computing have led to a proliferation of studies that focus on human activity recognition (HAR) utilizing inertial sensor data that consist of acceleration, orientation and angular velocity. However, the performances of such approaches are limited by the amount of annotated training data, especially in fields where annotating data is highly time-consuming and requires specialized professionals, such as in healthcare. In image classification, this limitation has been mitigated by powerful oversampling techniques such as data augmentation. Using this technique, this work evaluates to what extent transforming inertial sensor data into movement trajectories and into 2D heatmap images can be advantageous for HAR when data are scarce. A convolutional long short-term memory (ConvLSTM) network that incorporates spatiotemporal correlations was used to classify the heatmap images. Evaluation was carried out on Deep Inertial Poser (DIP), a known dataset composed of inertial sensor data. The results obtained suggest that for datasets with large numbers of subjects, using state-of-the-art methods remains the best alternative. However, a performance advantage was achieved for small datasets, which is usually the case in healthcare. Moreover, movement trajectories provide a visual representation of human activities, which can help researchers to better interpret and analyze motion patterns.},
  language  = {en}
}
@article{HampfNendelStreyetal.2021,
  author    = {Hampf, Anna and Nendel, Claas and Strey, Simone and Strey, Robert},
  title     = {Biotic yield losses in the Southern Amazon, Brazil},
  series = {Frontiers in plant science : FPLS},
  volume    = {12},
  journal   = {Frontiers in plant science : FPLS},
  publisher = {Frontiers Media},
  address   = {Lausanne},
  issn      = {1664-462X},
  doi       = {10.3389/fpls.2021.621168},
  pages     = {16},
  year      = {2021},
  abstract  = {Pathogens and animal pests (P\&A) are a major threat to global food security as they directly affect the quantity and quality of food. The Southern Amazon, Brazil's largest domestic region for soybean, maize and cotton production, is particularly vulnerable to the outbreak of P\&A due to its (sub)tropical climate and intensive farming systems. However, little is known about the spatial distribution of P\&A and the related yield losses. Machine learning approaches for the automated recognition of plant diseases can help to overcome this research gap. The main objectives of this study are to (1) evaluate the performance of Convolutional Neural Networks (ConvNets) in classifying P\&A, (2) map the spatial distribution of P\&A in the Southern Amazon, and (3) quantify perceived yield and economic losses for the main soybean and maize P\&A. The objectives were addressed by making use of data collected with the smartphone application Plantix. The core of the app's functioning is the automated recognition of plant diseases via ConvNets. Data on expected yield losses were gathered through a short survey included in an "expert" version of the application, which was distributed among agronomists. Between 2016 and 2020, Plantix users collected approximately 78,000 georeferenced P\&A images in the Southern Amazon. The study results indicate a high performance of the trained ConvNets in classifying 420 different crop-disease combinations. Spatial distribution maps and expert-based yield loss estimates indicate that maize rust, bacterial stalk rot and the fall armyworm are among the most severe maize P\&A, whereas soybean is mainly affected by P\&A like anthracnose, downy mildew, frogeye leaf spot, stink bugs and brown spot. Perceived soybean and maize yield losses amount to 12 and 16\%, respectively, resulting in annual yield losses of approximately 3.75 million tonnes for each crop and economic losses of US\$2 billion for both crops together. The high level of accuracy of the trained ConvNets, when paired with widespread use from following a citizen-science approach, results in a data source that will shed new light on yield loss estimates, e.g., for the analysis of yield gaps and the development of measures to minimise them.},
  language  = {en}
}
@phdthesis{Boeniger2010,
  author    = {B{\"o}niger, Urs},
  title     = {Attributes and their potential to analyze and interpret 3D GPR data},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus-50124},
  school      = {Universit{\"a}t Potsdam},
  year      = {2010},
  abstract  = {Based on technological advances made within the past decades, ground-penetrating radar (GPR) has become a well-established, non-destructive subsurface imaging technique. Catalyzed by recent demands for high-resolution, near-surface imaging (e.g., the detection of unexploded ordnances and subsurface utilities, or hydrological investigations), the quality of today's GPR-based, near-surface images has significantly matured. At the same time, the analysis of oil and gas related reflection seismic data sets has experienced significant advances. Considering the sensitivity of attribute analysis with respect to data positioning in general, and multi-trace attributes in particular, trace positioning accuracy is of major importance for the success of attribute-based analysis flows. Therefore, to study the feasibility of GPR-based attribute analyses, I first developed and evaluated a real-time GPR surveying setup based on a modern tracking total station (TTS). The combination of current GPR systems capability of fusing global positioning system (GPS) and geophysical data in real-time, the ability of modern TTS systems to generate a GPS-like positional output and wireless data transmission using radio modems results in a flexible and robust surveying setup. To elaborate the feasibility of this setup, I studied the major limitations of such an approach: system cross-talk and data delays known as latencies. Experimental studies have shown that when a minimal distance of ~5 m between the GPR and the TTS system is considered, the signal-to-noise ratio of the acquired GPR data using radio communication equals the one without radio communication. To address the limitations imposed by system latencies, inherent to all real-time data fusion approaches, I developed a novel correction (calibration) strategy to assess the gross system latency and to correct for it. This resulted in the centimeter trace accuracy required by high-frequency and/or three-dimensional (3D) GPR surveys. Having introduced this flexible high-precision surveying setup, I successfully demonstrated the application of attribute-based processing to GPR specific problems, which may differ significantly from the geological ones typically addressed by the oil and gas industry using seismic data. In this thesis, I concentrated on archaeological and subsurface utility problems, as they represent typical near-surface geophysical targets. Enhancing 3D archaeological GPR data sets using a dip-steered filtering approach, followed by calculation of coherency and similarity, allowed me to conduct subsurface interpretations far beyond those obtained by classical time-slice analyses. I could show that the incorporation of additional data sets (magnetic and topographic) and attributes derived from these data sets can further improve the interpretation. In a case study, such an approach revealed the complementary nature of the individual data sets and, for example, allowed conclusions about the source location of magnetic anomalies by concurrently analyzing GPR time/depth slices to be made. In addition to archaeological targets, subsurface utility detection and characterization is a steadily growing field of application for GPR. I developed a novel attribute called depolarization. Incorporation of geometrical and physical feature characteristics into the depolarization attribute allowed me to display the observed polarization phenomena efficiently. Geometrical enhancement makes use of an improved symmetry extraction algorithm based on Laplacian high-boosting, followed by a phase-based symmetry calculation using a two-dimensional (2D) log-Gabor filterbank decomposition of the data volume. To extract the physical information from the dual-component data set, I employed a sliding-window principle component analysis. The combination of the geometrically derived feature angle and the physically derived polarization angle allowed me to enhance the polarization characteristics of subsurface features. Ground-truth information obtained by excavations confirmed this interpretation. In the future, inclusion of cross-polarized antennae configurations into the processing scheme may further improve the quality of the depolarization attribute. In addition to polarization phenomena, the time-dependent frequency evolution of GPR signals might hold further information on the subsurface architecture and/or material properties. High-resolution, sparsity promoting decomposition approaches have recently had a significant impact on the image and signal processing community. In this thesis, I introduced a modified tree-based matching pursuit approach. Based on different synthetic examples, I showed that the modified tree-based pursuit approach clearly outperforms other commonly used time-frequency decomposition approaches with respect to both time and frequency resolutions. Apart from the investigation of tuning effects in GPR data, I also demonstrated the potential of high-resolution sparse decompositions for advanced data processing. Frequency modulation of individual atoms themselves allows to efficiently correct frequency attenuation effects and improve resolution based on shifting the average frequency level. GPR-based attribute analysis is still in its infancy. Considering the growing widespread realization of 3D GPR studies there will certainly be an increasing demand towards improved subsurface interpretations in the future. Similar to the assessment of quantitative reservoir properties through the combination of 3D seismic attribute volumes with sparse well-log information, parameter estimation in a combined manner represents another step in emphasizing the potential of attribute-driven GPR data analyses.},
  language  = {en}
}