Super-resolution land cover mapping aims at producing fine spatial resolution maps of land cover classes from a set of coarse spatial resolution class fractions derived from satellite information via, say, spectral unmixing procedures. Based on a prior model of spatial structure or texture that encodes the expected patterns of classes at the fine (target) resolution, this paper presents a sequential simulation paradigm for generating alternative super-resolution maps of class labels that are consistent with the coarse class fractions. Two modes of encapsulating structural information are investigated, one using a set of indicator variogram models, and one using training images. It is demonstrated that both such modes can be accommodated in the sequential simulation framework, and lead to super-resolution class maps that exhibit a variety of spatial patterns ranging from very simple to rather complex. A case study conducted using synthetic data illustrates the feasibility of the proposed geostatistical approaches for superresolution mapping for a binary (two-class) case over a relatively small regular grid.
展开▼
