Volume rendering is the term used for techniques that give a view of the entire volume, from a given viewing direction, to which all voxels are able to contribute.
Thus, the structure appears transparent or semitransparent, and all interior features appear overlapped. The simplest rendering of this kind is obtained by parallel projection (summation along rays in the viewing direction). However, more complicated schemes have been devised where interior densities are reassigned according to a classification scheme and gradient information is allowed to contribute.
The value of volume rendering is still controversial in many applications of image processing. The problem is that such rendering corresponds to no known experience in the physical world around us. Perhaps the closest experience is watching a jellyfish from a close distance and seeing its organs overlap as the animal moves. However, this example shows at the same time the value of watching a transparent structure in motion in order to appreciate the spatial relationships of its components. Stereo representation also helps in conveying these relationships, but experience shows that a presentation as a "movie" is by far the most powerful tool.
Somewhere in between surface and volume rendering falls the use of semitransparent surfaces in conjunction with solidly rendered interior density objects or atomic models. Surfaces represented as a contour mesh have an advantage over other semitransparent surface renderings in that they present sharply defined clues to the eyes in stereo presentations (see figure 6.7 in section 6.4). Seemingly more realistic and esthetic effects including reflections are, however, obtained by ray tracing (see for instance, Spahn et al., 2001a).
Was this article helpful?