“Volume Rendering on one RLE compressed data set by a new combination of Ray-Casting and Shear-Warp” by Schlosser, Hesser and Männer

  • ©Gregor Schlosser, Jürgen Hesser, and Reinhard Männer



Entry Number: 031


    Volume Rendering on one RLE compressed data set by a new combination of Ray-Casting and Shear-Warp



    In recent years, a growing demand to visualize ever larger (e.g. gigabyte) data sets poses the problem of handling large volumes and to render them in interactive times on a normal PC. One approach is based on lossy compression schemes. It compromises between high compression ratio, fast run-time random access and visualization quality. Our new approach based on shear-warp volume rendering, doesn’t have any visible artefacts due to quantization and after classification of the volume we achieve high compression ratios as well. This hybrid algorithm combines the advantages of image order and object-order algorithms. Further, we are able to preserve the benefits from Lacroutes original implementation. Like, volume rendering from run-length-encoded (RLE) data, alignment of volume and intermediate image (IIM) scanlines, only one object-order traversal of the volume, efficient and synchronized space leaping and early-ray-termination (ERT) using RLE data structures, preservation of the final warp step for efficient calculations. Nevertheless, we solved two problems found in the original implementation. Firstly, we removed the 3-fold redundancy of classified volumes (RLE data sets). Therefore it was necessary to develop a new rendering algorithm for one main viewing direction. We achieve better image quality (also preserving the Shannon-Nyquist theorem), using a ray-casting approach within 4 RLE scanlines and new data structures for efficient ERT. Secondly, we increased the rendering quality for the other two main viewing directions by extending the discrete volume grid to continuous space by Hermite polynomials.


ACM Digital Library Publication:

Overview Page: