While volume rendering is becoming more widely used ill medical application
s, it is still difficult to generate a good-quality image interactively wit
hout expensive hardware when the size of the dataset is quire large. For in
teractive rendering of a large dataset, we present an acceleration method,
image-space bounding surface. Using an image-space bounding surface, the is
osurface ray casting is accelerated by avoiding unnecessary volume traversa
ls. Image-space bounding surface can be interactively handled by polygon-re
ndering hardware even in a conventional personal computer. Two optimization
techniques, LF-minmax map and memory bricking., are also employed to effic
iently render isosurfaces. This paper also shows that thr algorithm can be
extended to multiple isosurfaces rendering. Thr experimental results show t
hat the algorithm generates a good-quality image of a large dataset interac
tively on a standard PC platform (C) 2000 Published by Elsevier Science Ltd
.