Distance Transformations in Hexagonally Digitized Images (Avstandstransformationer i Hexagonalt Digitaliserade Bilder)

摘要

Distance transformations is an important tool for processing binary digital images. Recently pseudo-Euclidean, or weighted, distance transformations have been receiving quite a lot of attention. Such transformations use only local operations and preferably integer arithmetic. In the paper pseudo-Euclidean distance transformations on the hexagonal grid are investigated. Optimal distance transformations for different neighborhood sizes are derived, as well as good integer approximations of them. There are several reasons why the hexagonal grid is important: it is a reasonable approximation of the human vision grid; it is more suited to natural scenes in low resolution than the rectangular grid; and it avoids the annoying 4-neighbor/8-neighbor problem.