2003 Volume 16 Issue 9 Pages 484-491
This paper proposes a fast method, called cyclic optimal multi-step method, for searching nearest target object in the road network under the spatial database, in which both target objects and road network are indexed by R-trees. This method consists of filtering and refinement steps, which run cyclically. By using this method, the nearest search begins from selecting a target with the shortest straight-line, measured from a source, as the candidate for path search and then determines a search region which contains the selected candidate target and road-based path. To implement this method, a priority queue is introduced to record nodes of R-tree and candidate objects, and also a variable which indicates the radius of search region is done to record the mediate results in searching the minimum path. In this case, the key by which the elements on the queue are to be ordered is the distance from the source object. Since in our method the path search process is limited only inside the search region and moreover only the object located at the head of the queue is considered, the search process is greatly accelerated.