Our objective is to design a numerically efficient algorithm for interpolation of nominal data in large data cubes. More specifically, the problem is to determine a complete data cube from raster images corresponding to parallel layers of the data cube such that the complete data cube interpolates all given layers. The given raster images display discrete geoobjects, i.e. nominal data which may be the result of some classification of initial data. We suggest nominal interpolation by virtue of interpolation of the corresponding multivariate membership function with radial basis functions to recover 3d geoobjects the sections of which interpolate the series of raster images.
展开▼