This title appears in the Scientific Report :
2002
Please use the identifier:
http://hdl.handle.net/2128/2584 in citations.
Betrachtungsabhängige Visualisierung von Höhenfeldern mittels Wavelet-basierter Methoden zur Detailkontrolle
Betrachtungsabhängige Visualisierung von Höhenfeldern mittels Wavelet-basierter Methoden zur Detailkontrolle
The performance of modern graphic workstations has increased a lot during the last years, now supporting the display ofthousands offilled polygons at interactive frame rates using normal PC hardware . With growing graphic performance model complexity has also increased, so that in many graphical app...
Saved in:
Personal Name(s): | Delonge, F. (Corresponding author) |
---|---|
Contributing Institute: |
Zentralinstitut für Angewandte Mathematik; ZAM |
Imprint: |
Jülich
Forschungszentrum Jülich GmbH Zentrlbibliothek, Verlag
2002
|
Physical Description: |
VIII, 74 S. |
Dissertation Note: |
Aachen, Fachhochsch., Abt. Jülich, Dipl., 2002 |
Document Type: |
Diploma Thesis |
Research Program: |
Betrieb und Weiterentwicklung des Höchstleistungsrechners |
Series Title: |
Berichte des Forschungszentrums Jülich
4020 |
Subject (ZB): | |
Link: |
OpenAccess |
Publikationsportal JuSER |
The performance of modern graphic workstations has increased a lot during the last years, now supporting the display ofthousands offilled polygons at interactive frame rates using normal PC hardware . With growing graphic performance model complexity has also increased, so that in many graphical applications the data size still sustancially exceeds the capacity of the graphics hardware . This problem becomes particullary apparent when considering height fields, i.e. height values with respect to a regular grid: The unsimplified display of the triangulated surface of a realistic model which consist of up to 100 million gridpoints takes roughly about 20 seconds an a state-of-the-alt graphic PC. However, interactive data handling requires at least 25 frames per second, resulting in a maximum display time of 40ms per single image. Nevertheless, in order to visualize large data sets interatively, methods to simplify the scene complexity are needed. Although these methods must reduce the number of triangles representing the surface by orders of magnitude, their impact an the image quality should be barely noticeable . In this work, a well scalable framework of algorithms based an wavelet transformations with view dependant error metrics is presented, allowing for the neccessary reduction of data points in real time. Also, consecutive aspects of triangultion and rendering are discussed. |