Research On3D Visualization Based On Nonphotorealistic Rendering Technology

In the research area of visualization, it is still an unsolved problem that how tomake the structural information which is contained in the volume data expressedaccurately and automaticly. The traditional visualization methods based on physicoptical models are mainly trying to generate photorealistic images, but they arehardly to present the important structural information of the volume data. In thevisualization, the use of the non-photorealistic rendering techniques can flexiblycontrol the appearances of the rendering results, overcoming the lack ofphotorealistic rendering methods, and showing the structural features of the volumedata easily.The volume rendering methods are widely used in the visualization area. Thispaper presents a boundary enhancement method based on the classification of theopacity values and the gradient values of the voxels. The method first divides thevoxels into two regions, the boundary region and the non-boundary region, based onthe gradient values. And the opacity values of the voxels in the boundary region willbe increased, in the opposite the opacity values of the voxels in the non-boundaryregion will be reduced. Meanwhile, use the k-means method to classify the voxels’original opacity values into different categories, and make sure that each categoryhas a distinguished enhancement effect. The experiments prove that this method canmake a good performance. In this paper we also improve the original contourenhancement method and the tone shading method. We introduce some thresholdvalues in order to control the rendering results flexibly. And the experiments provethat the improved methods performing better than the original method. We implementthe ray casting algorithm using GPU accelerating, which can increase the renderingspeed, achieving the real-time rendering, and making the observation moreconvenient.