Summary

Interactivity is crucial for efficient exploration and analysis of volume data. Complex data sets require careful and frequent tuning of visualization parameters to obtain meaningful visualization results. The specification of a proper transfer function [23,27,29,34], i.e., the assignment of optical properties to data values within the volume is a complex task which benefits greatly from immediate visual feedback by interactive rendering of the volume. The main obstacle for interactive volume rendering is simply the amount of data to be processed for generating an image from a volumetric data set. Typical volume sizes in medicine range from $256^3$ voxels for MR data to $512^2\times 2000$ voxels for data acquired with recent multi-detector CT scanners. For a straight-forward approach, this would mean shading and compositing 16-500 million voxels for each single image - a tough task even for multi processor hardware. Simple straight forward implementations of volume rendering are only competitive in terms of performance if directly implemented in hardware - like the VolumePro (vp500) volume rendering board from Real Time Visualization [48].

Within this work, a novel, purely software-based solution to interactive rendering of volumetric data is presented, which is able to deliver interactive frame rates even on low-end hardware. The approach is also well-suited for use in networked environments due to a compact data representation. Several distinguishing features make the presented method a fast and flexible solution to interactive, software-based volume rendering for low-end hardware:

• preprocessing: during a preprocessing step, voxels which potentially contribute to a visualization result are identified.
• voxel enumeration: possibly contributing voxels are extracted from the volume and stored in a derived data structure, which is basically a list of individual voxels.
• compact representation: the extracted voxels are well-suited for efficient compression and can be transmitted over a network for visualization at a remote computer at very low cost.
• voxel ordering: the extracted voxels can be ordered in a way which is optimized for rendering using specific compositing modes and visualization parameter settings.
• fast rendering: A fast shear/warp-based rendering [28] is used to project the extracted voxels.
• object awareness: if segmentation information is available, extracted voxels can be assigned to individual objects. Visualization parameters can be defined on a per-object basis, allowing to individually adjust opacity and color transfer functions, shading models, and even compositing modes, without much impact on rendering performance.

The voxel extraction approach can be seen as a hybrid approach between direct volume rendering, which directly operates on the original volume data, and approaches like marching cubes [33], which derive a polygonal representation of objects within the volume for rendering. On one hand, only a secondary data representation which represents the volume is used for rendering - the list of potentially contributing voxels. On the other hand, the voxel data within this data structure is just a space-efficient storage representation for a sparsely populated volume.

mailto:mroz@cg.tuwien.ac.at.