Summary: | 碩士 === 大同工學院 === 資訊工程學系 === 84 === The speed of spatial query affects not only the efficiency
offeature searching but also the processing time of various
geographic data manipulation and analysis. Thus, constructing a
spatial indexfor fast region queries is very important for the
development of a successful Geographic Information System (GIS)
system. There are many technical papers on spatial index and
region queries,for example, R-tree, R+-tree, KD-tree, and
quadtree etc. In this thesis,an adaptation of quadtree, called
multiple storage quadtree, is described as a spatial index
strucutre. We will introduce the multiple storagequadtree to
support dealing with polygon intersection problems. In the
polygon intersection problems, we usually use a minimumbounding
rectangle (MBR) to enclose a polygon as an approximation of
thepolygon. If an MBR overlaps some quad region of the multiple
storage quadtree, we should insert this object to the
corresponding leaf quad.If the amounts of objects in a leaf quad
overflows some threshold value, the leaf quad will be
recursively decomposed into four new leaf quadsuntil no leaf
quads overflows. This process keeps the characteristic thatthe
objects which spatially close to each other will be in one quad
leaf. Using this spatial relationship, we can reduce a lot of
unnecessary segmentintersection tests in polygon intersection
problems. In this thesis, we make a little improvement, called
monotonous cut, of cutting each polygon object into several
polylines which are monotonouslyincreasing or decreasing along
the y-direction. Each polyline is enclosed in an MBR and then
inserted into the multiple storage quadtree. By
introducingmonotonous cut, we can use a merge comparison
approach to reduce a quantity of segment intersection tests and
we also maintain the spatial relationshipeach polygon object.
The advantage of this structure is that by using MBRs and
multiple storage quadtree, we can reduce a lot of segment
intersection tests according to the spatial relationship between
polygon objects. The disadvantage of this structureis that we
may make multiple examination of the same objects in a single
query.
|