|
Scientific Visualization
| Issue Year: | 2012 |
| Quarter: | 2 |
| Volume: | 4 |
| Number: | 2 |
| Pages: | 21 - 56 |
|
| Article Name: |
APPROACH TO OPTIMIZATION OF GPU-ALGORITHM FOR THE VOLUME RAYCASTING METHOD TO LEVEL OF USING IN A VIRTUAL DISSECTING TABLE |
| Authors: |
N. Gavrilov (Russia), V. Turlapov (Russia) |
| Address: |
N. Gavrilov
gavrilov86@gmail.com
Lobachevsky state university of Nizhny Novgorod, Russia
V. Turlapov
vadim.turlapov@cs.vmk.unn.ru
Lobachevsky state university of Nizhny Novgorod, Russia |
| Abstract: |
In this article we describe approaches we have implemented to improve the visualization quality and performance of our volume rendering engine which can be used in The Virtual Dissection Table for high-resolution medical datasets exploration, applied in Stanford University's teaching process. We describe particular details of our GPU-based Ray Casting algorithm that deals with bricked datasets, i.e. divided into blocks. We show that such division increases the benefits of such optimization strategies as early ray termination and empty space skipping, performing high-level early ray termination and bricks’ automatic bounding boxes fitting correspondently. To reduce visual artifacts we use the random ray start position generation and further frames accumulation, pre-integrated rendering approach, we also present adaptive sampling rate and ray step subdivision methods. The quality is also improved by on-the-fly tri-cubic filtering during the ray casting process. We use the Deep Shadow Mapping for the shadows rendering to improve the intuitive visual perception of the surface depth and shape. We outlined the visualization performance in terms of frame rates for different rendering techniques using different block sizes. |
| Language: |
Russian |
|
|
|
