A directional path distance model for raster distance mapping

Abstract

This paper proposes a directional model for weighted path distance mapping on raster data structures. The approach that is used to map the shortest path distance from a set of source cells to any background cell incorporates both non-directional and directional weights on the links between adjacent cells. Nondirectional weights allow users to model impedance in terms of the traditional cell based friction costs used in raster distance mapping. Directional weights allow users to model non-symmetric travel costs incurred in traveling through value-gradients in continuous fields such as elevation, temperature or density, or incurred when traveling through a prevailing flow field. The model has been implemented as a cell-based raster distance mapping tool with user selectable directional factor functions, using both Dijkstra's and the Bellman-Ford's shortest distance algorithms. Two experimental results are included.