CPU-based speed acceleration techniques for shear warp volume rendering
Multimedia Tools and Applications
| Hi-index | 0.00 |
Shear–warp volume rendering has the advantages of a moderate image quality and a fast rendering speed. However, in the case of dynamic changes in the opacity transfer function, the efficiency of memory access drops, as the method cannot exploit pre-classified volumes. In this paper, we propose an efficient algorithm that exploits the spatial locality of memory references for interactive classifications. The algorithm inserts a rotation matrix when factorizing the viewing transformation, so that it may perform a scanline-based traversal in both object space and image space. In addition, we present solutions to some problems of the proposed method, namely inaccurate front-to-back composition, the occurrence of holes, and increased computation. Our method is noticeably faster than traditional shear-warp rendering methods because of an improved utilization of cache memory. Copyright © 2005 John Wiley & Sons, Ltd.