Data Structures for Spatial Grids
Spatial grids can be represented in a number of ways: as an array of numbers, a list of pixels, a level set, a topographic surface, a Markov chain, a matrix, a parametric approximation, a multiresolution pyramid, or as a multiscale tree. Each of these representations makes the grid amenable to specific types of processing. We illustrate the benefits of each of these representations on the problem of extracting the most populated cities in North America from a population density grid. Along the way, we explore Radial Basis Function, projection pursuit, and Gaussian Mixture Models.
KeywordsGaussian Mixture Model Spatial Grid Projection Pursuit Negative Amplitude Topographical Representation
- Broomhead DS, Lowe D (1988) Multivariable functional interpolation and adaptive networks. Complex Syst 2:321–355Google Scholar
- Klooster S, Potter C, Kumar V, Tan P-N, Steinbach M (2003) Discovery of climate indices using clustering. In: Ninth ACM SIGKDD international conference on knowledge discovery and data mining, Washington, DC. ACM, p 446Google Scholar