本文提出了一种基于Delaunay三角形化图象“表面”描述的灰度图象编码方法。利用Delaunay三角形化的良好特性,仅编码由少数象素组成的特征图象就可以几乎不失真地恢复出描述得到的近似模型。对常规Delaunay三角形化的改进避免了由于退化产生的三角形连接错误。整个编译码方案在压缩比为33倍时仍具有较好的恢复图象质量。 This paper presents a gray-level image coding method based on the Delaunay triangulation image “surface” description. Using the good features of Delaunay triangulation, the approximation model described can be recovered with little distortion by coding only a feature image composed of a few pixels. Improvements to conventional Delaunay triangulation avoid triangle connection errors due to degradation. The entire coding and decoding scheme still has a better image quality at a compression ratio of 33 times.