Accelerating Morphological Analysis with Graphics Hardware.
Matthias Hopf, Thomas Ertl
Abstract
Matthias Hopf, Thomas Ertl
Abstract
Direct volume rendering is a common means of visualizing three-dimensional data nowadays. It is, however, still a very time consuming pro-cess to create informative and visual appealing images. Semi-automatic volume analysis proce-dures as morphological operators are a promis-ing approach to improve the overall visualization cycle. However, these operators need quite some time for computation, reducing their usefulness for interactive visualization. Modern graphics hardware, on the other hand, has all necessary functions for doing hardware based morphological filtering. As the problem is mainly memory bandwidth bound, a solution based on graphics hardware can significantly re-duce computation time in the filtering step, as the graphics hardware typically has much broader and faster memory paths. When using graphics hardware for mathe-matical computations, accuracy is usually quite a problem. However, morphological operators map so well onto the graphics pipeline, that no loss of accuracy occurs at all. 1
OpenAlex reports 24 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Direct volume rendering is a common means of visualizing three-dimensional data nowadays. It is, however, still a very time consuming pro-cess to create informative and visual appealing images. Semi-automatic volume analysis proce-dures as morphological operators are a promis-ing approach to improve the overall visualization cycle. However, these operators need quite some time for computation, reducing their usefulness for interactive visualization. Modern graphics hardware, on the other hand, has all necessary functions for doing hardware based morphological filtering. As the problem is mainly memory bandwidth bound, a solution based on graphics hardware can significantly re-duce computation time in the filtering step, as the graphics hardware typically has much broader and faster memory paths. When using graphics hardware for mathe-matical computations, accuracy is usually quite a problem. However, morphological operators map so well onto the graphics pipeline, that no loss of accuracy occurs at all. 1
Key concepts: Graphics hardware, Graphics pipeline, Computer science, Rendering (computer graphics), Graphics, Real-time computer graphics, Computation, Visualization