· · · Institutional Repository

Oud warenhuis GLANS aan het kerkplein staat leeg en is ideaal om een nieuw publiek gebouw in te huisvesten; het plein, het gebouw en de ruimte rondom bieden plek voor vernieuwing en levendigheid. GLANS wordt een ontmoetingsplek in de stad door de verlenging van het stedelijk interieur in het gebouw en door veel diverse functies in GLANS bij elkaar te brengen; een ontmoetingsplek met als programma o.a. een bibliotheek en sportfuncties. Het herontwerp brengt zo dynamiek in het centrum. Ik wil uiteindelijk een verandering maken door een nieuw pubiek gebouw in Paramaribo te ontwerpen waar ontmoeting en kennis centraal staan.

Planar partitions (subdivisions of the plane into polygonal areas) constitute one of the most important data representations in GIS. They are used to model concepts as varied as land use, administrative units, natural features and cadastral parcels, among many others. However, since polygons are often stored separately, different errors and inconsistencies are introduced during their creation, manipulation (both manual and automatic) and exchange. These come in the form of invalid polygons, gaps, overlaps and disconnected polygons, which severely hampers their use in other software. Existing approaches to solve this problem usually involve polygon repair using a list of constraints, and complex planar partition repair operations performed on a planar graph. However, these have many shortcomings in terms of complexity, numerical robustness and difficulty of implementation. Moreover, they leave many invalid cases untouched. To solve this problem, a novel method to validate and automatically repair planar partitions has been developed. It uses a constrained triangulation of the polygons as a base, which being by definition a planar partition, means that only relatively simple operations are needed to ensure that the output becomes valid. Point locations are maintained throughout the process, while fully automatic repair is possible using customisable criteria. This approach is also extensible to individual polygons, is capable of handling a larger variety of cases and has good performance compared to existing alternatives; all of this with numerical robustness and maintaining topological consistency throughout. In order to analyse, test and improve the developed algorithms, and encourage further development, a fast and efficient implementation has been written in C++, which has been tested with several large data sets and compared with other available software, regarding both performance and functionality. This prototype is able to successfully repair planar partitions of more than 100,000 polygons. It is also open source and freely available on the GDMC website (http://www.gdmc.nl/).

With the increase of the availability of large-scale geographic data set and the rise of widespread computer networks, such as the Internet, the need has arisen to be ableto transfer this data by means of these networks. The networks form the basis for a Geographic Information Infrastructure (GII), in which data users, data providers and data producers are connected with each other.There also exists the need to offer this data on several scales to end users, for example to get an overview of an area first. Because the geographical information are now sent by the computer networks and large-scale geographic information brings many data, data reduction must take place. This is to prevent that sending of the information takes too much time. Generalization of geographical information is a possible means to let this reduction take place.Generalization is the selection and simplification of detail appropriate to the scaleand or purpose of the map. The appliance of generalization demands that choices mustbe made with respect to which geographical objects are selected and simplified and how this selection and simplification must take place. Moreover, also the surroundings of the objects to be generalized, are often taken into account in the generalization process, which makes that the complete process even requires more time. This way, the complete process can not be carried out in real time.Earlier, reactive data structures, in which geographical information is stored in the computer with several levels of detail, have been proposed as a solution to allow the use of generalized large-scale geographical information in real time. So far, these data structures were using redundancy with respect to geometry. For this reason a new conceptual model has been developed, where a number of existing data structures have been combined into two new data structures, namely, the GAP face tree and the GAP edge forest (described in Van Oosterom, 2005). The complete structure is termed tGAP structure, inwhich tGAP stands for topological Generalized Area Partitioning.The tGAP structure has not been theoretically verified, nor implemented or tested. Therefore, the objective of this research is to theoretically verify the data structures and to test the data structures considering requirements such as loading time and storage capacity.To reach the objective literature study has been performed in the field of generalization, database management systems and the data structures. Moreover, a prototype has been built, with which the data structures have been implemented in a mainstream database management system (DBMS) with spatial data types. Literature study has shown that generalization is a key issue in the complete process of obtaining and processing geo-information and that using reactive data structures is asuitable option to offer the results of generalization within a GII in real time.The implementation of a prototype has shown, that it is possible to implement thedata structures in a mainstream DBMS. The data structures are implemented in Oracle Spatial, the DBMS, and by means of Apache, a web server, opened up to Google Earth, a geographical viewer. The data structures in the prototype make it possible to view thegeographical data interactively within the viewer independent from the size of the area to be loaded.The final conclusion must be, that with some workarounds and with some changes tothe proposed conceptual model, it is possible to implement the model as described in (Van Oosterom, 2005). With an implementation it becomes possible to show geographical data on a variable number of detail levels and the implementation shows that the data structures can provide the desired data reduction within a GII in real time.

This paper introduces the concept of a space-scale partition, which we term the space-scale cube – analogous with the space-time cube (first introduced by Hägerstrand, 1970). We take the view of ‘map generalization is extrusion of 2D data into the third dimension’ (as introduced by Vermeij et al., 2003). An axiomatic approach formalizes the validity of the partition of space in three dimensions (2D space plus 1D scale). Furthermore the paper provides insights in how to: 1. obtain valid data for the cube, 2. obtain a valid 2D polygonal map at variable scale from the cube and 3. which other possibilities the cube brings for obtaining maps having different map scales over their domain (which we term mixed-scale maps).

One of the simplest methods to construct a 3D city model is to extrude building footprints to obtain \block-shaped" polyhedra representing buildings. While the method is well-known and easy to implement, if the 2D topological relationships between the footprints are not taken into account, the resulting 3D city models will not necessarily be topologically consistent (i.e. primitives shared by 3D buildings will be duplicated and/or intersect each others). As a result, the model will be of little use for most applications, besides visualisation that is. In this paper, we present a new extrusion procedure to construct topologically correct 3D city models. It is based on the use of a constrained triangulation, is conceptually simple, and offers great exibility to create city models in different formats (e.g. CityGML or a surface-based representation). We have implemented the procedure, tested it with real-world datasets, and validated it.

One of the simplest methods to construct a 3D city model is to extrude building footprints, to obtain "block-shaped" buildings. While the method is well-known and easy to implement, if the topological relationships between the footprints are not taken into account, the resulting city models will not necessarily be topologically consistent. As a result, the model will be of little use for most applications, besides visualisation that is. In this paper, we present a new extrusion algorithm to construct topologically correct 3D city models. It is conceptually simple and permits us to create city models in different formats (e.g. CityGML). We have implemented the algorithm, tested it for the creation of the model of our university campus and validated it by constructing the constrained Delaunay tetrahedralization.

In this paper, we will give a description of a design of a fat client and study some implementation issues, to use the tGAP structures for progressive data streaming. This is an experiment to validate the theory of Haunert et al. (2009). Furthermore this theory is extended and a solution is proposed to make the progressive data streaming more cache-friendly by means of a Fieldtree.

This paper presents the problem of the current separate treatment of levels of detail in city models. We propose a solution, detail the main principles, and present our initial results on the approach. We conclude with work in progress and explain the benefits of our approach.

A method for generating a vario-scale visual representation of n-dimensional objects (together forming a space partition) is presented comprising the steps of - generating a higher and a lower detailed n-dimensional object representation each comprising digital data representing objects by zones in said n-dimensional object representations, said zones being delimited by (n-1)-dimensional boundaries having at least one boundary segment, - positioning the higher and lower detailed object representation in an (n+1)-dimensional space, having in addition to the dimensions of the n-dimensional object representations an additional dimension, wherein the higher and the lower detailed n-dimensional object representations are assigned a first and a second value for said additional dimension respectively, - constructing an (n+1); -dimensional object representation by creating trans-scale boundary segments between mutually corresponding boundary segments in the higher detailed and the lower detailed n-dimensional representation, - determining an intermediate n-dimensional representation by calculating a cross-section between an n-dimensional slicing object and the constructed trans-scale boundaries.

Planar partitions—full tessellations of the plane into non-overlapping polygons—are frequently used in GIS to model concepts such as land cover, cadastral parcels or administrative boundaries. Since in practice planar partitions are often stored as a set of individual objects (polygons) to which attributes are attached (e.g. stored with a shapefile), and since different errors/mistakes can be introduced during their construction, manipulation or exchange, several inconsistencies will often arise in practice. The inconsistencies are for instance overlapping polygons, gaps and unconnected polygons. We present in this paper a novel algorithm to validate such planar partitions. It uses a constrained triangulation as a support for the validation, and permits us to avoid different problems that arise with existing solutions based on the construction of a planar graph. We describe in the paper the details of our algorithm, our implementation, how inconsistencies can be detected, and the experiments we have made with real-world data (the CORINE2000 dataset).

This paper deals with efficient data management of variable scale vector data. Instead of pre-building a collection of data sets on different scales, we create an index structure on the base data set (largest scale data) that enables us to extract a map at exactly the right scale the moment we need it. We present both the classic version of the tGAP (topological Generalized Area Partitioning) data structure for storing our variable scale map, as well as an ameliorated version, both based on topological concepts. We prove that the classic structure needs in a worst case scenario O(e2) edges (with e the number of edges at largest scale). In practice we observed up to a factor 15 more edges in the variable scale data structure. The tGAP structure has been optimized to reduce geometric redundancy, but the explosion of additional edges is due to the changing topological references. Our main achievement finds its roots in the reduction of the number of edge rows to be stored for the ‘lean’ version (by removing the topological referential redundancy of the classic tGAP), which is beneficial both for storage and transfer. We show that storage space for the data set plus the index, is less than twice the size of the original data set. The ‘lean’ tGAP, as the classic tGAP, offers true variable scale access to the data and has also improved performance, mainly due to less data communication between server and client.

This paper presents the first true vario-scale structure for geographic information: a delta in scale leads to a delta in the map (and smaller scale deltas lead to smaller map deltas until and including the infinitesimal small delta) for all scales. The structure is called smooth tGAP and its integrated 2d space and scale representation is stored as a single 3d data structure: space-scale cube (ssc). The polygonal area objects are mapped to polyhedral representations in the smooth tGAP structure. The polyhedral primitive is integrating all scale representations of a single 2d area object. Together all polyhedral primitives form a partition of the space-scale cube: no gaps and no overlaps (in space or scale). Obtaining a single scale map is computing an horizontal slice through the structure. The structure can be used to implement smooth zoom in an animation or morphing style. The structure can also be used for mixed-scale representation: more detail near to user/viewer, less detail further away by taking non-horizontal slices. For all derived representations, slices and smooth-zoom animations, the 2d maps are always perfect planar partitions (even mixed-scales objects fit together and form a planar partition). Perhaps mixed-scale is not very useful for 2d maps, but for 3d computer graphics it is one of the key techniques. Our approach does also work for 3d space and scale integrated in one 4d hypercube.

The use of geo-information is changing by the advent of new mobile devices, such as tablet-pc's that harness a lot of computing power. This type of information is more and more applied in mainstream digital consumer products, in a net-centric environment (i.e. dissemination takes place via the Internet) and the advances in mobile hardware also have changed the way people can interact with the geographic information at hand, compared to `old-fashioned' paper maps. However, current state-of-the-art solutions for storing, maintaining and disseminating digital maps still mimic the analogue map-series concept in the sense that for every map scale in the serie (e.g. 1:25K, 1:50K, 1:250K) a different digital copy with independent data is kept and maintained at the producers site. The challenge of this work was to get to a representation of the real world with gradually changing level of detail, instead of representations with discrete levels of detail (organised in multiple, independent layers, each layer representing only one resolution level). Vario-scale data structures try to avoid this redundancy of the geometric description of the map by storing references to composing map elements of the highest level of detail for any other element of a lower level of detail. An example of variable-scale data structures are the tGAP data structures. In addition to the geometry and references, an importance value for every object is stored and based on this importance value different representations (where the level of detail is gradually changing) can be derived on the fly from these structures according to the needed level of detail. The overall aim of this research has been to investigate variable-scale geo-information, by defining theoretical underpinnings of vario-scale geo-information and improving the initial tGAP structures. The objective we had with this research is expressed in the main question, which was formulated as: How can we realise improved vario-scale geo-information having minimal redundancy? The overall outline of the research design draws heavily upon the paradigm of design research. In an iterative fashion we performed theory building, prototype developments and experiments with real world data sets. Over the course of this research, we have made the following main contributions to the design of a vario-scale geo-information environment. We have: - formalised the concept of variable-scale data as a conceptual 3D model (the space-scale cube, SSC), where 2D space and 1D scale is integrated; - shown for the tGAP data structures how minimal data redundancy can be obtained when applying a merge operation, how to perform a parallel simplification of lines, without introducing unwanted topological errors and proposed a split operation, for which it was analysed what the impacts are on the designed data structures; - shown how to derive a 2D map from the structures with a particular number of objects, as well as investigated progressive data streaming; - proposed an improved way of generating data so that even smoother graphic transitions can be derived for visualisation. The main conclusions that can be drawn from these contributions: - With the concept of the proposed space-scale cube (SSC) we have formalised what vario-scale vector data entails. In a sense, the improved design of the tGAP data structures can be seen as a lossless encoding of the data that is captured for a ssc; - To make vario-scale geo-information operational, we need specific generalisation operations. These vario-scale generalisation operations should be designed carefully to be able to give guarantees on the amount of data to be stored and output topologically consistent vario-scale data; - Although the improved tGAP structures are capable of providing a smooth zooming end user experience, we still store and visualise discrete steps -- albeit smaller and more local than is common with current state of the art solutions. Therefore we proposed how smoothness of the vario-scale data can be improved (where the smooth SSC taking a small step in scale leads to a small change in the 2D derived map). A novelty of this approach is that, as it is one integrated space-scale partition, using a non-horizontal slice plane leads to a valid, mixed-scale planar partition: this is useful for use in 3D computer graphics (far away from an observer having less detail than close by). Although this research has generated some knowledge for a vario-scale environment, it also paves the way for future research. The main recommendations for future work are: - Investigate how to deal with very large data sets that do not fit in main memory (during the generalisation process or during visualisation) deserves attention; - The smooth encoding of the SSC has the same building challenge as the classic tGAP with respect to applying the right sequence of generalisation operators (remove or merge, collapse or split, simplify) to obtain maps with sufficient cartographic quality; - Another point for further research is the smooth interactions: it is of importance to know how users perceive these. The same holds for mixed-scale slices (in a 3D world); - Focus of this research has been mostly on obtaining and viewing vario-scale data. Performing analysis with vario-scale data is another interesting aspect that deserves attention, e.g. vario-scale data could be of help in data integration; - Investigate how to make the structures dynamic: currently the tGAP structure (including the new smooth variant) is a static structure and has to be re-built if the source data changes. Being able to perform incremental updates (partially re-generalising data for a new situation) would be beneficial if the data volume increases. Related to this is higher dimensionality of smooth, vario-scale data (e.g. 3D data) leading to integrated 5D data management (integrating dimensions of space (2D or 3D), time (updates, 1D) and scale (level of detail, 1D).