Projects / Programmes
Visualization of landscape on distributed computer systems
| Code |
Science |
Field |
Subfield |
| 2.07.03 |
Engineering sciences and technologies |
Computer science and informatics |
Programming technologies - software |
| Code |
Science |
Field |
| P170 |
Natural sciences and mathematics |
Computer science, numerical analysis, systems, control |
Visualization, parallel algorithms, distributed systems, simulation
Researchers (1)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
21555 |
PhD Marjan Šterk |
Computer science and informatics |
Head |
2007 - 2008 |
69 |
Organisations (1)
Abstract
Computer visualization of landscape nowadays has uses in various applications ranging from displaying simulations of natural phenomena to virtual urban planning and virtual tourism. Many applications find uses in close systems, such as the ones in state institutions use, while the applications targeted to the general public have recently been emerging as well, e.g., Google Earth (http://earth.google.com). However, data and computational complexity of the visualization algorithms limit these applications either in performance, image quality or model size.
The proposed project will speed up the visualization through algorithmic improvements and, mainly, through distributed execution on distributed systems and grid systems.
The main focus of algorithmic improvements will be the use of meshless shape functions instead of the standard B-splines. The conversion of the point-based three-dimensional model to a triangular mesh will be omitted, lessening the computational complexity and simplifying model modifications.
The visibility and required image quality on different parts of the model depends on the current viewing point and angle, complicating the distributed execution. A scheduler will thus be developed that will distribute the workload to the individual computers dynamically, on-the-fly. To efficiently distribute the execution and avoid high communication overhead, the data starting data distribution will also have to be well chosen.
The project will collect all theoretical results and implement them in a software visualization framework, which will be based on an open-source license. A test application based on the framework will also be developed.
Significance for science
The large number of papers published lately about visualization of huge datasets, advanced visualization algorithms, and scheduling of requests among distributed servers, speak for international interest in these topics. We have shown that it is possible to serve a large number of visualization clients using a relatively simple redundant server infrastructure. The performed analysis of request scheduling algorithms and the developed client-side algorithm, which removes the need for a single access point and server-to-server communication, solve some open problems in these areas and can also serve as a basis for future research on request scheduling algorithms.
Despite the discovery that modeling the terrain using meshless base functions does not have any significant advantages, the improvement of the MLS base function stability will make them more easily applicable to other problems, such as solution of partial differential equations. The implementation of MLS base functions on graphic processors is also important for these problems, as this implementation can provide a significant speed-up in these problems and is not in any way bound to visualization.
The open-source visualization framework XGIS is also important scientifically, as it represents a stable platform for execution of repeatable and comparable experiments and can thus support future research of the topics. Also, the framework is an instructive case of quick application of scientific discoveries to practical problems.
Significance for the country
Geographical information systems and especially three-dimensional landscape visualization are extremely important for a developing country like Slovenia. Such tools can improve urbanistic and infrastructure planning. Various parts of Slovenia are also subject to different kinds of natural disasters, whose frequency and intensity is increasing with the climate changes. Three-dimensional visualization can improve our preparation for such events as well as the emergency response when such a disaster indeed strikes. Systems developed in Slovenia have an edge in adaptation to specific local requirements and especially in maintenance and dependability assurance during disastrous events. Distributed storage and automatic rescheduling to replicated servers using the developed scheduling algorithm enables the XGIS-based applications to guarantee high availability even in case of damaged communication infrastructure. The ability to use locally stored data is important in case of complete failure of the communication infrastructure. The XGIS framework has already been used for visualization of flooded areas within the scope of the targeted research project Interactive visualization of flooded areas to support emergency activities (in Slovene, CRP Interaktivna vizualizacija poplavnih območij za podporo interventnemu ukrepanju).
The fact that the XLAB company is developing a new product Gaea+, which used the results of this project as the first prototype, is important for Slovenian economy. Also, the published scientific results will contribute to the international academic visibility of Slovenia.
Most important scientific results
Final report,
complete report on dLib.si
Most important socioeconomically and culturally relevant results
Final report,
complete report on dLib.si
