Social Sensing (2020)
Perception of social-event-induced human behavior from geotagged social media data
Contact: Dr.-Ing. Ruoxin, Zhu
The population coverage of public transit systems is an important indicator of public transit accessibility. Traditionally, the assessment of transit catchment areas is mainly focused on walking as the access mode. The recent emerging dockless shared bikes are widely used for connecting with public transit systems and provide new chances to expand the population coverage of public transit. This project aims to assess how dockless shared bikes could expand the transit catchment areas using massive bike trajectories. To achieve this aim, the project has three objectives: 1) proposing a fast method to generate network-based transit catchment areas for non-motorized transport; 2) proposing methods to measure biking distances from high-detailed bike trajectories; 3) conducting cases studies to evaluate the effectiveness of the proposed methods and discussing policy implications for the planning of public transit and dockless shared bikes.
Modeling of Public Transit Accessibility Driven by Spatial Movement Data (2020)
Contact: Dr. Ing. Diao Lin
This dissertation focuses on investigating the bike-metro integration using spatial movement data and supporting a systematic assessment of accessibility to public transit. Specifically, it serves three research objectives: 1) exploration of biking distances at individual transit stations from trajectory and smart card data, 2) investigation of transit catchment area to raise the public awareness of the transit accessibility at a general level, and 3) inspection of accessibility constrained by crowdedness at a fine-grained level.
Approaching a collective place definition from street-level images using deep learning methods (2019)
Contact: Dr. Ing. Hao Lyu
This work addresses the challenge of understanding a place in GIScience by investigating visual and spatial (semantic) property in voluntarily collected images with deep learning methods. Two place representations are proposed to unify these properties in a probabilistic perspective of understanding places. The proposed computational models which are based on comparative learning and variational autoencoder are proved to be able to learn the probabilistic place representations from image data.
Visualizing Uncertainty in Reasoning - A Bayesian Network-enabled Visual Analytics Approach for Geospatial Data (2019)
Contact: Dr. Ing. Ekaterina Chuprikova
The growing importance of data-driven science and advances in computational capacity offer new opportunities for the analysis and visualization of geospatial and heterogeneous data. This dissertation addresses the challenges of analytical reasoning under conditions of uncertainty when working with spatial data. It serves three research objectives: (1) to evaluate the feasibility of the Bayesian Network in representing conditional dependencies among heterogeneous spatial data; (2) to implement visual analytics scenarios that can demonstrate human-data discourses; (3) to build a prototype of a Bayesian Network-enabled visual analytical system dedicated to geospatial data classification tasks.
Conflation of Road Networks from Digital Maps (2016)
Contact: Dr. Ing. Andreas Philipp Hackelöer
Road Network Conflation is concerned with finding accurate mappings between geographical structures of road networks. This thesis establishes Road Network Conflation within a taxonomy of georeferencing methods and classifies and compares common approaches to the problem. A novel approach called Iterative Hierarchical Conflation (IHC) is introduced, which systematically accounts for the resolution of ambiguities. The IHC is evaluated using several samples from digital maps of different vendors. Results show that the IHC offers advantages especially in terms of correctness, speed and complexity.
Labeling Spatial Trajectories in Road Network using Probabilistic Graphical Models (2016)
Contact: Dr. Ing. Jian Yang
Labeling spatial trajectories, such as map matching, activity recognition, can ease the utilization of the imprecise and semantic poor spatial trajectories for location-aware applica-tions. This thesis studies the problem from a unified perspective using map matching and taxi status inference. Comprehensive probabilistic models are learned from the training data using a chain structure graphical model with feature selection, which are tested to be effective and feasible on a real world dataset.
Visual Analysis of Large Floating Car Data - A Bridge-Maker between Thematic Mapping and Scientific Visualization (2016)
Contact: Dr. Ing. Linfang Ding
This thesis aims to bridge the gaps between thematic mapping and scientific visualization and to achieve their synergetic effects for the visual analysis of big data. Firstly, a systematical comparative study of thematic cartography and scientific visualization is conducted. The study shows that these two disciplines reveal different visual analytical levels and are mutually complementary. Next, extensive experiments of visually analyzing massive real-world taxi floating car data (FCD) have been carried out. The experiment results demonstrate that the techniques from thematic mapping and scientific visualization can strongly support users to win insight into the movement data.
Event Cartography: A New Perspective in Mapping (2016)
Contact: Dr.-Ing. Nina Polous
In this research, the concept of mapping goes beyond the principle of mapping an object as a conceptual geographic entity with a distinct spatial, temporal and attributive identity. The main goal is to present a conceptual model for managing geo-knowledge which handles real world dynamisms. It uses a generic event-oriented perspective to implicitly represent causal relationships among different components of a Spatio-Temporal Information System. From this new perspective, the objects in space and time are considered merely as information elements of the events, which are connected to other event elements through internal or external processes.
Dynamics of spatially extended phenomena (2014)
Contact: Dr.-Ing. Stefan Peters
This thesis focuses on the visual exploration of a specific type of moving geoobjects, namely the spatially extended objects or phenomena. Visual analytical approaches are developed and implemented to study the dynamics of the spatio-temporally evolving polygons. The lightning data are chosen as a real-world case. In addition to a generic concept for the movement analysis of spatially extended objects, the thesis put forward a number of synchronized cartographic and non-cartographic visual analytical approaches for the clusters.
Concise Image Maps - A Design Approach (2014)
Contact: Dr.-Ing. Christian E. Murphy
An image map is a composition of remote sensing imagery and cartographic symbolisation. This work revisits the concept of image maps and shows that the conventional two-tiered structure can be extended by the concise image map design, according to which a differentiated visual hierarchy can be established. Therefore, design strategies are developed that address the radiometric design of raster imagery in the same manner as the graphical design of map symbols. User tests evaluate several concise image map design strategies that prove to be more effective and user friendly.
A Congruent Hybrid Model for Conflation of Satellite Image and Road Database (2013)
Contact: Dr.-Ing. Jiantong Zhang
This thesis is devoted to the conflation of two heterogeneous data sources - road vectors and geo-referenced images. The contributions of the Congruent Hybrid Model (CHM) include:1) a linear feature extraction approach, which consists of Elastic Circular Mask (ECM) algorithm and the Genetic Algorithm (GA)-based grouping approach;2) a Sparse Matching Algorithm (SMA) approach; and 3) a performance evaluation of two transformation functions. The CHM model can be used directly in the geovisualization applications, and with some modifications it is also suitable for the classical georeferencing problem.
Nicht-Photorealismus in der Stadtmodellvisualisierung für mobile Nutzungskontexte (2013)
Contact: Dr.-Ing. Mathias Jahnke
Die Visualisierungen dreidimensionaler Stadtmodelle erschließen bisher nicht das Potential, der kombinierten geometrisch semantischen Informationsdarstellung. Der aus dem Nicht-Photorealistischen-Rendering bekannte Ansatz der Informationsreduzierung durch Abstraktion lässt sich mit kartographischen Gestaltungsprinzipien kombinieren und liefert unter Einbeziehung von Nutzerpräferenzen neue Formen der Geo-Informationskommunikation auf der Basis von Stadtmodelldaten.
Enrichment of routing map and its visualization for multimodal navigation (2012)
Contact: Dr.-Ing. Yueqin Zhu
This thesis is dedicated to a novel approach for the design of customized routing maps which demonstrate the route along with the fast rendering of a right amount of routing-relevant information that matches the cognitive capacity of the user on route. Given a route with mono- or multimodality, the proposed approach first estimates the salience of individual mapping objects by combining the passive salience from internal characteristic of spatial data and the active salience from the participant. The achieved results are visualized in multi-scale routing map with a dynamic labeling algorithm. The proposed approach is implemented as web-based services. Its feasibility has been verified in several experiments.
Kartenicons im interkulturellen Vergleich (2011)
Contact: Dr. rer. nat. Stephan Angsüsser
Das übergeordnete Ziel dieser Arbeit ist ein Beitrag zur kulturvergleichenden Kartographie. Ausgehend von der Annahme, dass Unterschiede zwischen den Karten verschiedener Kulturen bestehen, ist eine bestimmte Art von Kartenzeichen als Beispiel herangezogen worden. Dabei handelt es sich um Kartenicons, deren Gemeinsamkeit normalerweise in ihrer geringen Größe und weitgehenden Isoliertheit besteht. Zu deren Analyse wurde auf der Basis bestehender Ansätze ein neues Zeichenmodell entwickelt. Für jedes der 1016 Kartenicons (540 deutsche und 476 chinesische) wurden 8 Attribute bestimmt, deren Vergleich schließlich 25 wesentliche Unterschiede zwischen den beiden Länderauswahlen ergab. Um diese zu erklären, wurde versucht, kulturelle Eigenschaften heranzuziehen. Schließlich führte dies zur Formulierung von 38 Hypothesen über mögliche Beziehungen zwischen Eigenheiten der deutschen bzw. chinesischen Kultur und Eigenheiten der in diesen beiden Ländern produzierten Kartenicons.
Kartographische Anreicherung von Gebäudefassaden mit thermalen Bilddaten (2011)
Contact: Dr.-Ing. Holger Kumke
Die vorliegende Arbeit behandelt die visuelle Anreicherung thermaler Daten auf Gebäudefassaden imstädtischen Raum. Wärmestrahlung, messtechnisch als bildhafte Thermogramme erfasst, liefern neuenicht sichtbare Informationen über den Gebäudezustand und dienen als Rohdaten für diekartographische Aufbereitung zu thermalen Fassadenkarten für planare 2D wie auch kartenverwandteräumlich virtuelle 3D Darstellungen. Aus einer Kombination der bekannten Gestaltungsregeln und derneuen kartographischen Perspektive entstanden system-unabhängige thermale Fassadenkarten, dieals Anregung, Denkanstöße und Ausblick auf temperaturbezogene Darstellungsformen im städtischenRaum und Grundlage für weitere Forschungsarbeit dienen sollen.
Distributed geo-services based on Wireless GIS - a case study for post-quake rescue information system (2011)
Contact: Dr.- Ing. Yimei Liu
A useful application of Wireless GIS is the handling of natural disasters such as earthquakes. This thesis is dedicated to the construction and implementation of a post-quake rescue information system based on open source data and software programs. The emphasis is laid on the assessment of losses in the disaster area, estimation of collapsing buildings and trapped population, and efficient transmission of all the rescue-relevant information. The realized workflow of using open source data and software programs to develop distributed geo-services for rescue purposes is independent of official data sources, therefore, flexible enough to react on emergency situations.
Generalization of Road Network for an Embedded Car Navigation System (2011)
Contact: Dr.- Ing. Hongbo Gong
Automatic map generalization serves to reduce the amount of data to speed up the mapping process or to ensure the legibility of small scale maps. This thesis deals with the task of automatic selection of road networks for the application of visualization and route planning in an embedded car navigation system. Based on an intensive analysis of the embedded system in terms of storage capacity, the display screen and the necessary computing power in real time, two special constraints - connectivity and network density – are introduced. A concept for the semantic-driven path selection for the map display and the optimal route planning is developed and implemented with test data from Germany and China.
Data Model and Algorithms for Multimodal Route Planning with Transportation Networks (2011)
Contact: Dr.- Ing. Lu Liu
Determining a best route in highly developed complex transportation networks is not a trivial task, especially for those who are unfamiliar with the local transportation system. Multimodal route planning that aims to find an optimal route between the source and the target of a trip while utilizing several transportation modes is essential to intelligent multimodal navigation services. Although the task originates from the field of transportation, it can be abstracted as a general form independent of the domain-specific details on the underlying data model and algorithms. This research work is therefore dedicated to a general approach of modeling the multimodal network data and performing optimal path queries on it. The weighted digraph structure can well represent the fundamental static networks. For each mode, there is one corresponding mode graph. These graphs constitute the Multimodal Graph Set as a key component of the overall multimodal network data model. In comparison with the traditional mono-modal problem, another key component necessary in the modeling of multimodal route-planning problem is mode-switching actions. In this work, such actions are described by Switch Points which are somewhat analog to plugs and sockets between different mode graphs. Consequently, it is possible to plug-and-play a Multimodal Graph Set by means of Switch Points. On the basis of the multimodal network data model, the multimodal route-planning problem is categorized into two types and formalized as the multimodal shortest path problem on the Plug-and-Play Multimodal Graph Set. It turns out that the solutions for these two types of problem are equivalent if the input mode list for the first type is transformed into its matrix expression. When applying the general multimodal route-planning approach to a specific application domain, a rule-based inferring process is necessary to determine whether a mode sequence is reasonable or not. Performance evaluations on the integrated navigation dataset have verified the efficiency of the proposed approach.
Driver Behaviors on Different Presentation Styles of Traffic Information (2010)
Contact: Dr.- Ing. Masria Mustafa
Road traffic information has been one of the important elements for traffic information systems. The information can be found on the road where Variable Message Sign (VMS) and other platforms have been widely utilized for such application. These overwhelming majorities of traffic information sources provide real-time traffic information, aiming at helping drivers make better decisions on choosing a correct traffic route on the basis of current traffic state. However, it is an unusual sight to view the full scenarios of the system with the view from the drivers themselves. Therefore, smartly presenting the information to the road user has a potential to support the road user to receive and interpret the information in a more effective way. Traffic information presented in different styles may enhance the attractiveness of the information itself. Considering this, we design this study to find out whether there is a refined relationship between the specific presentation style and the driving behavior. As a starting point, this study tested the assumption that the probe vehicle could provide reliable and sufficient amount of data that could represent travel time information which then can be transformed into various presentation style. VISSIM 5.0 is used to generate travel time data on a hypothetical network. Average travel time on links are analyzed for various percentages of probe vehicles and compared to the ‘true’ average travel time using ‘bootstrapping’ technique. ArcGIS designed for use by transportation professionals to display the results of travel time provided by probe in a more understandable visual fashion (color coded design). Later, user testing upon preference of the drivers towards different types of traffic information presentation style is conducted. A picture is often cited to be worth a thousands words and, for some tasks it is clear that a visual presentation such as map is dramatically easier to be used than other textual or spoken description. Visual displays become even more attractive to provide orientation or context, to enable selection of regions and to provide dynamic feedback for identifying change such as dynamic traffic congestion map. In what concerns the visual information, systems can present information using graphics, symbols or even written messages. A stated preference user test is conducted and questionnaires with different types of traffic information presentation style are distributed to the respondents. An underlying question is basically about whether and how the presentation styles of traffic information affect the driver in making their decision. The study addresses a wide range of alternative styles of in-vehicle traffic information as well as stationary information in different driving scenarios (stop and go and congested). The analysis is carried out which contains various trip variables, including route selection characteristics, travel purpose and actual observable traffic conditions en route such as level of congestion, variables pertaining to the information to which is being displayed and also psychological factors based on personal attributes and the experience of the individual drivers. It is assumed that drivers are influenced by these variables and factors of making decisions whether to acquire and refer to traffic information in choosing their route. Our results revealed that in case of in-vehicle information, presentation style of traffic information does not play a significant role for driver’s behavior. As to the preference of presentation style, ‘map with detail building’ came out to be the highest rank. The main reason for this preference is the presence of the buildings which provides additional orientation information. Different behavior patterns could be observed when confronted with more realistic situations. Our observations demonstrate that the drivers are more likely to divert their route only in rush trip. In case of stationary information, again, we found no evidence that presentation style of traffic information does play a significant role for driver’s behavior. As to the preference of presentation style, ‘combination of graphic and text information’ came out to be the highest rank. Our observations demonstrate that the drivers are more likely to divert their route only in rush trip and congested route.
Integration of time-dependent features within 3D city model (2010)
Contact: Dr.- Ing. Hongchao Fan
This thesis presents an object-oriented event-state spatiotemporal data model for storage and management of both semantic and geometric changes of 3D building objects in a city. The data model is mainly composed of two parts: an event model that describes events happened to building objects; and a hierarchical spatial data model that describes 3D geometries and semantics of building objects including their valid time span. In this way, histories of building objects are modeled. The data model can be “double indexed” by events happened to objects and by objects involved in events. Correspondingly, queries can be triggered by both events and objects. On this base, a set of spatiotemporal queries are proposed. The spatiotemporal data model proposed in this work combines the advantages of event-based model and object-based spatiotemporal data model. On one hand, dynamic processes are modeled as events with their types/classes, locations, time points/durations, modes of the processes, and the involved city objects. On the other hand, the life of an object is represented by a time-ordered sequence of its states and the dynamic processes indicating how the object changes from one state to another. The approach of storing events and city objects separately reveals a number of benefits: (i) the multiple storage due to n-to-m relations among events and objects are avoided, (ii) the spatiotemporal data model is double-indexed. Events and 3D objects can be queried independently and efficiently. In addition, the proposed spatiotemporal data model takes the hierarchy and inherent relations between events and objects into account, so that both events and 3D objects can be represented at different levels of detail.
Methods and Implementations of Road-Network Matching (2009)
Contact: Dr.-Ing. Meng Zhang
Data matching is one of the fundamental measures that helps make different data sets interoperable. This thesis is devoted to a new contextual matching approach for road networks. This automatic matching process is based on the Delimited-Stroke-Oriented algorithm and flanked by three assistant methodologies: matching guided by 'structure', matching guided by 'semantics', and matching guided by 'spatial index': Being supported by the extendable delimited strokes, network-based matching and the three assistant methodologies, the contextual matching approach is able to handle geometrical, topological and semantic information in a large matching environment and provide a considerably improved matching performance in terms of ‘automatic matching rate and certainty', 'high computing speed', and 'robustness and generic nature'. Due to its large potentials of enriching mega data sets, the contextual matching approach is being commercialized.
Attention-Guiding Geovisualisation: A cognitive approach of designing relevant geographic information (2008)
Contact: Dr. rer. nat. Olivier Swienty
It is a delicate task to design suitable geovisualisations that allow users an efficient visual processingof the depicted geographic information. In digital era, such a design task is subject tothree major challenges: the ever growing amount of geospatial data at various levels of detail,the diversified applications of that data, and the continuously expanding range of display sizes.These challenges are guided by the same cognitive scope. Users face an increasing level ofcognitive workload that has a substantial impact on decision-making while processing complexvisual environments.This work tends to enhance the visualisation of relevant geographic information by proposing aconceptual framework for the development of attention-guiding geovisualisation. The mainchallenge is to stimulate a users decision-making and to reduce the cognitive workload by providinghigh responsiveness in specific visual brain areas that are involved in visual geographicinformation processing. Based on theories and research findings in GIScience and cognitiveneuropsychology the research basis of this work is formed by combining utility and usabilityissues of system engineering.The relevance of information is considered as an utility criterion and its cognitively adequatevisualisation as an usability criterion of a system’s acceptability. To enhance utility, irrelevantinformation is separated from relevant information by implementing relevance as a filter. Toenhance usability the design of attention-guiding geovisualisation is adapted to internal visualcharacteristics of visual information processing.Based on the internal structure of visual information processing and biological mechanismsinvolved in visual attention, appropriate cognitive principles and a design methodology arepresented and applied to pixel-based remote sensing satellite image and vectorised maps. Apre-evaluation with a computational attention-model serves as a knowledge base for designingvectorised attention-guiding geovisualisations that are evaluated with a paper and pencil testand the eye-movement recording method.The evaluation results reveal that the proposed attention-guiding design approach significantlyenhances visual geographic information processing and contribute to the overall acceptabilityof geographic information systems and geovisualisations that are needed for fast and accuratedecision-making processes.
Recognition of 3D Settlement Structure for Generalization (2005)
Contact: M.Sc., M.Tech. Jagdish Lal Raheja
This thesis aims at recognizing 3D settlement structures for automatic generalization, an innovative extension to 2D and their simplification based on scale-spaces. The recognition procedure has been divided into three levels namely micro, meso and macro and is based upon individual buildings, buildings in neighborhood and buildings at cluster level having similar properties such as settlement blocks as well as psychophysically perceived groups. Any of these three levels of structure recognition demands that comprehensive information about the buildings should be known a-priori. These buildings, simple as well as complex, are recognized using an Artificial Neural Network (ANN) in a bottom-up approach. It starts with recognizing ground plans of buildings and which in turn, along with other information, are used to recognize different roof types and finally entire buildings are recognized in a similar way. After building recognition, their structure description has been studied in detail, which gives rise to various measurable parameters of individual as well as buildings in neighborhood. These parameters not only characterize individual building but also many spatial relations among them. Structure recognition at clustered level is studied next and it involves the recognition of group of buildings as a whole. The human visual system can detect many clusters of patterns and significant arrangements of image elements. Perceptual grouping refers to the human visual ability to extract significant image relations from lower-level primitive image features without any knowledge of the image content and group them to obtain meaningful higher-level structure. Various perceptual grouping principles have been applied to identify these clusters of groups. After a comprehensive study of structure recognition, their findings are then applied to 3D generalization. Among the various generalization algorithms such as aggregation, displacement, simplification, exaggeration, typification, aggregation is chosen here as it almost uses most of the results from structure recognition. Various constraints resulting from spatial relations have been already found in 2D aggregation. However, unlike in 2D, where there is only one view, the third dimension leads to many additional views and these different views become the source of additional conflicts. Apart from various views, color, texture and other small parts (window, chimney, balcony etc.) of the building also add to the existing constraints. Various additional rules have been obtained based upon these constraints. These rules along with the results of structure recognition have been used to trigger the aggregation operation.
Mobile Cartography - Concepts for Adaptive Visualisation of Spatial Information on Mobile Devices (2004)
Contact: Dr. rer. nat. Tumasch Reichenbacher
This PhD project developed the theoretical and conceptual framework of mobile cartography. The main focus is on the elaboration of adaptive methods for the visualisation of spatial information for mobile usage, i.e. on mobile devices (e.g. PDA). The starting point for the adaptation is the mobile user, his activities and goals, as well as the situation these three are placed in. Usage scenarios helped to implement a prototype geo-service for mobile users based on open-standard formats such as XML, GML, and SVG, which serves as a proof of concept.