| 摘要 |
A ray tracing method based on a directional coherence map comprises partitioning image data into blocks of pixels, with each block including edge pixels and interior pixels. The blocks are individually processed by first computing a discrepancy for each of a plurality of directions of the block. For a given direction, the discrepancy corresponds to a sum of differences of opposing edge pixels oriented along the direction. The direction having a least discrepancy is then selected as an edge direction for the block. Next, a bilinear interpolation is performed along the edge direction to compute the remaining pixels of the block. The approximate image is produced when all of the blocks have been processed.
|
