An algorithm for shortest raster distance in Euclidean space with obstacles

Abstract

The application of the shortest raster distance algorithm in Euclidean space without obstacle is unsuitable and inaccurate in real geographical environment. A new efficient algorithm is presented in this paper which aims to obtain shortest Euclidean distance in a space with multiple point sources and polygonal obstacles based on raster data model. First the turning points on the routes are defined as the tangential points on the tangent lines from point sources to obstacles' convex. Then, search the raster cells according to certain rules of turning points and connect them to build the routes which present the shortest distances from sources to the destinations bypassing obstacles. The key techniques of this algorithm are to detect whether a straight route is crossing an obstacle, to define turning points, and to make approximation to the destination through turning points. The results show that the algorithm is highly practical, stable and accurate.