The shear-warp volume rendering algorithm is one of the fastest algorithms for volume rendering, but it achieves this rendering speed only by sacrificing interpolation between the slices of the volume data. Unfortunately, this restriction to bilinear interpolation within the slices severely compromises the resulting image quality. This paper presents the implementation of pre-integrated volume rendering in the shear-warp algorithm for parallel projection to overcome this drawback. A pre-integrated lookup table is used during compositing to perform a substantially improved interpolation between the voxels in two adjacent slices. We discuss the design and implementation of our extension of the shear-warp algorithm in detail. We also clarify the concept of opacity and color correction, and derive the required sampling rate of volume rendering with postclassification. Furthermore, the modified algorithm is compared to the traditional shear-warp rendering approach in terms of rendering ...
Jürgen P. Schulze, Martin Kraus, Ulrich Lang,