基于Delaunay三角网生长法的并行图像插值方法

摘要

This paper proposes a parallel image interpolation method based on the algorithm of incremental construction of Delaunay triangulations. This method diminishes the search scope of the optimal circum-circle through eight-point neighbourhood options, and accelerates the speed of the edge update by adopting points-storing based Delaunay edges link table. The parallel Delaunay image interpolation on a cluster environment is implemented by splitting strategy. The method has the characteristic of little memory, and can solve the problem of image interpolation for large data sets.%提出一种基于Delaunay三角网生长法的并行图像插值方法.该方法通过八邻域备选点减小了最优外接圆搜索范围,并采用了基于点存储的Delaunay边链表,加快了边更新速度,通过划分策略实现了机群环境下的并行图像插值.该方法占内存小,可以解决大数据量的图像插值问题.