Spatial-frequency characteristics of satellite images of homogeneous landscapes with high spatial resolution received in panchromatic spectral channel are investigated. Four types of terrain are considered. they are forest, cutting area, meadow and settlement. The autocorrelation matrices for chosen fragments of images are calculated. The spatial autocorrelation surfaces are created on their basis. Analytical approximations of these surfaces are selected and also the spectral densities and values of correlation radiuses are calculated. It is shown that the autocorrelation functions of one-dimentional brightness cut sections of images greatly differ from corresponding cut sections of autocorrelation surfaces.