Description |
Scientific visualization techniques center around graphical data representations as opposed to symbolic or numerical designations. As a technique in scientific visualization, volume rendering allows scientists, engineers, and researchers, and physicians to view and manipulate a three dimensional scalar field. For my honors thesis, I implemented a volume rendering algorithm and added my code into a scientific programming environment for computational steering. This incorporation allows the visualization technique to be used along with other popular data processing methods. After implementing the basic method for rendering regular grids presented by Levoy, I modified and optimized Levoy's ray casting approach by changing the original lighting model, employing a faster sampling method, introducing a variable for faster "preview" image generation, and porting the code to a parallel machine. In addition, I provided a flexible and user-friendly interface to the program. I allowed my module to communicate with a three-dimensional viewer to allow the rendered volume to be viewed along with an isosurface (generated by another visualization technique). Although I was able to modify and optimize the Levoy method, I realized that the original approach provided little room for significant improvement; therefore, I explored other volume rendering techniques. Beneficial enhancements for my module are discussed in papers featuring the shear-warp factorization method for regularly grided data, and the direct volume rendering method based on the scan procedure and hierarchy for irregular or multiple grids. |