Experimental evaluation of filter effectiveness (extended abstract)
Lixian Han, Hongjun Zhu, Jianwen Su · 2000
Evaluation of a spatial join typically has two steps: a filter step that applies the join on approximations of spatial objects and a refinement step that determines the actual intersection of two spatial objects. In this paper, we study the effectiveness of filters with different approximations including minimum bounding rectangles, rectilinear approximations, minimum 5-corner bounding convex polygons, and convex hulls. We compare the effectiveness of filters by comparing the number of “false hits” in the filter result using these approximations. We analyze the impact of parameters such as approximation quality and aspect ratios of input polygons on the filter effectiveness.