The doctoral dissertation addresses the problem of out-of-core point cloud visualization with a point-based rendering method. Our main goal is the development of data structures and algorithms for fast and quality out-of-core visualization of unprocessed point clouds. A survey of point-based rendering methods is presented, followed by an introduction of a new data structure based on the sequential point trees. It is realized as an octree with nested modified sequential point trees and allows efficient out-of-core rendering of unprocessed point clouds. It also supports visibility culling and rendering in different levels of detail. A rendering method which renders points as scaled, rotated and shaded Gaussian splats is presented. It provides better depth feedback and fills the gaps between points, both of which are very important especially for editing operations. Editing operations that enable selection, insertion and deletion of points in out-of-core point clouds are developed. Finally, the presented data structures and algorithms are analyzed and hypotheses are confirmed.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 264986368This doctoral thesis presents a domain-specific modeling environment for creating measurement procedures that are performed on measurement systems. Designing of domain-specific modeling language is introduced, and its effective use in contrast to general-purpose languages is explained. Although engineers usually use just the abstractions from the measurement domain without creating any overlap with the programming code, additional tools that add to their comprehension are more than welcome. For the efficient detection of modeling errors, a debugging tool is presented. In the context of this tool, different techniques, such as execution modes, steps, breakpoints, animation, variable views, and stack traces, which assist with localizing and eliminating mistakes during the debugging of the model, are explained. During the process of developing measurement systems, testing of measurement procedures is very important. For that purpose, this thesis defines an upgrade process of the above-mentioned language that allows for creating test cases for hardware as well as software. The thesis also presents a tool for the restructuring of measurement procedures. In the context of this tool, modeling procedures that are specific to the measurement domain, but which do not affect the behaviour of the measurement system, are presented. At the end of this doctoral thesis, a method for the automatic creation of new constructs on the basis of pre-modeled measurement procedures is presented. With this method, we have raised the level of abstraction and therefore make faster programming for the domain-experts possible. A domain-specific modeling environment is included in the DEWESoft package which is one of the most efficient measurement tools available. The modeling environment is presented with the help of practical examples from the automotive and space industry.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 265161472This paper presents an innovization design principle for procedural tree model of woody plants (trees) reconstruction by multi-objective optimization. Reconstruction of a parameterized procedural model from imagery is addressed by a multi-objective differential evolution algorithm, which evolves a parametrized procedural model by fitting a set of its rendered images to a set of given projected images using bi-objective comparisons made on a pixel level of the images. The use of multi-objective approach gives the decision maker a chance to select the final resulting model, and helps determine the optimization criteria tradeoff weights for later production.
B.03 Paper at an international scientific conference
COBISS.SI-ID: 16116758In the field of DSM (Domain-Specific Modeling), a popular software development technique, the metamodel plays an important role as it represents a schema definition of the syntax and static semantics to which a model conforms. Although model comparison techniques are widely researched and presented in the community, metamodels are higher level than models resulting in the failure of directly applying current model comparison tools. In this paper, we present a modeling tool named MMDiff for metamodel comparison, detecting the matching and differencing between two metamodels.
B.03 Paper at an international scientific conference
COBISS.SI-ID: 15917334Cloud Computing is an approach that provides computation and storage services on-demand to clients over the network, independent of device and location. In the last few years, cloud computing became a trend in information technology with many companies that transfer their business processes and applications in the cloud. Cloud computing with service oriented architecture has contributed to rapid development of Geographic Information Systems. Open Geospatial Consortium with its standards provides the interfaces for hosted spatial data and GIS functionality to integrated GIS applications. Furthermore, with the enormous processing power, clouds provide efficient environment for data intensive applications that can be performed efficiently,with higher precision, and greater reliability. This paper presents our work on the geospatial data services within the cloud computing environment and its technology. A cloud computing environment with the strengths and weaknesses of the geographic information system will be introduced. The OGC standards that solve our application interoperability are highlighted. Finally, we outline our system architecture with utilities for requesting and invoking our developed data intensive applications as a web service.
B.03 Paper at an international scientific conference
COBISS.SI-ID: 16418838