ABSTRACT
This chapter focuses on concepts, techniques, and applications in image processing that can be found in the overlap of data structures and statistical analysis. It provides a coherent description of procedures that has been found efficient for resolution-dependent processing in remote sensing image analysis. The chapter highlights some of the important implications of regular multiresolution decompositions of digital images using hierarchical data structures. The spatial statistical characteristics of spatial data have been extensively used in limiting, or determining the accuracy of estimating “missing values”. Typically, estimation is based on simplifying assumptions about the overall covariance structure of the scene. The Haar-representation of images is a flexible, computationally efficient tool in simulation. It is particularly attractive in remote sensing image analysis, because the variance partitioning function can be easily characterized based on images of finer resolution.
