Computer Vision and Image Understanding
Change Analysis in Spatial Data by Combining Contouring Algorithms with Supervised Density Functions
PAKDD '09 Proceedings of the 13th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining
Discovery of interesting regions in spatial data sets using supervised clustering
PKDD'06 Proceedings of the 10th European conference on Principle and Practice of Knowledge Discovery in Databases
On the shape of a set of points in the plane
IEEE Transactions on Information Theory
Analyzing the composition of cities using spatial clustering
Proceedings of the 2nd ACM SIGKDD International Workshop on Urban Computing
| Hi-index | 0.00 |
In general, it is desirable to have automatic tools that identify places in spatial data and to describe their characteristics, creating high-level summaries for spatial datasets which are valuable for planners, scientists, and policy makers. In this paper, we present a methodology that identifies a set of places based on a user-defined notion of interestingness and then identifies the scope of each place. A spatial clustering approach is used for the first step. For the second step, polygons are used as models to describe the scope of a place---the spatial area the place occupies. A 2-step methodology is introduced to compute a set of polygons for a set of places with each space being characterized by the set of objects which occupy the particular space. In the first step, an algorithm called LDTR is introduced that tightens the convex hull of a set of spatial objects by removing larger triangles of its Delaunay triangulation, obtaining an initial polygon for each place. Next, a post processing algorithm PolyRepair is introduced that tightens polygons further by reducing the overlap between the generated polygons; the algorithm gives preference to tightening polygons that have a lot of overlap with other polygons as the goal is to keep polygon tightening to a minimum. Finally, the two novel algorithms are demonstrated and evaluated for an urban computing benchmark.