| 摘要 |
A method and apparatus for providing real-time processing of voxels and real-time volume visualization of objects and scenes in a highly parallel and pipelined manner includes a three dimensional (3-D) skewed memory, two dimensional (2-D) skewed buffers, 3-D interpolation and shading of data points, and signal compositing. The method and apparatus implement ray-casting, a powerful volume rendering technique. Viewing rays are cast from the viewing position into a cubic frame buffer and beams of voxels, which are parallel to the face of the cubic frame buffer, are accessed. At evenly spaced sample points along each viewing ray, each sample point is tri-linearly interpolated using values of surrounding voxels. Central differences of voxels around the sample points yield a gradient which is used as a surface normal approximation. Using the gradient and the interpolated sample values, a local shading model is applied and a sample opacity is assigned. Finally, ray samples along the ray are composited into pixel values and provided to a display device to produce an image.
|
