Application of spatial filter theory to kriging |
| |
Authors: | James R. Carr |
| |
Affiliation: | (1) Department of Geological Sciences, Geological Engineering Division, University of Nevada-Reno, 89557 Reno, Nevada |
| |
Abstract: | Data-processing requirements for remotely sensed, digital images include spatial filtering to suppress image noise, enhance edges/contacts, and improve image clarity. Spatial filter theory demonstrates that the addition of a high-pass filtered image to a low-pass filtered image yields the original digital image. Application of this principle in kriging can be accomplished by using the same covariance matrix to solve for two weighting vectors to yield a result analogous to low- and high-pass filtering. The addition of kriged estimates calculated using both weighting vectors is analogous to summing high-, and low-pass filtered digital images. This modified method of kriging yields estimates associated with less smoothing compared to ordinary kriging. Statistical moments of original sample data are better preserved through estimation by this method. |
| |
Keywords: | point-spread function spatial filtering kriging variogram digital image intrinsic random functions conditional simulation |
本文献已被 SpringerLink 等数据库收录! |
|