Depth-Based Feature Enhancement for Volume Visualization

Direct volume rendering (DVR) is established as a powerful tool for volume visualization, which accumulates the color and opacity contributions by means of a simple light transport model. However, the mapping from data attributes to the optical properties is defined by transfer functions, the design of which is always a challenging and time-consuming task, even for expert users. In order to build informative images from original volume datasets without specifying intricate transfer functions, we present in this paper a depth-based feature enhancement visualization technique that could provide all features along the viewing ray at once, even with a simple linear transfer function. Once the accumulated opacity overflows, our approach resorts to an adaptive depth-based weighting to reduce the accumulated opacity value for adjusting the contribution of each voxel in the final pixel. To improve the visual perception of interesting features, such a modulation would be further enhanced when local maximum structures are encountered along the viewing ray. In addition, we provide a focus and context interaction and achieve a depth-based clipping operation to help users distinguish the order of internal structures. We conduct experiments on several volumetric datasets, and more structural information is presented and features of interest are largely enhanced in our rendering results, which further demonstrates the effectiveness of our proposed method.