Image coding using irregular subsampling and Delaunay triangulation

The author proposes an image coding algorithm that effectively combines irregular subsampling and Delaunay triangulation. A set of sampling points in the image is selected in an incremental manner and used to construct a Delaunay triangulation. Each triangular region is then approximated using linear interpolation. The coding scheme is based on run-length coding (RLC). The incremental selection of the sampling points takes into account the image contents. The results show that the algorithm is robust and superior to JPEG at low bit rates of 0.20 bpp and 0.10 bpp.