INTERZONAL ANALYSIS OF IMAGES USING MULTIDIMENSIONAL-MATRIX STRUCTURES

Images of a multilayer structure are fairly often encountered in pract ical problems of the recognition and analysis of subjects. These inclu de multizonal video photographs obtained by remote sensing in the visi ble part of the spectrum and in the infrared band, as well as radar im ages. The most important information in such photographs lies in the c hanges between the images of the same subject obtained in different sp ectral bands. An analysis of the nature of these changes enables one t o extract additional information on the state and characteristics of t he objects being observed, for example, vegetation on the Earth's surf ace, ice, stretches of water etc. The second group of multilayer image s consists of cartographic information, superimposed on maps of the di stribution of the chemical composition of the cell, geobotanical, biol ogical-soil and other characteristics of the Earth's forests or tilled fields. The functional agreement between these characteristics in spa tial coordinates enables one to obtain not only a more-accurate estima te of the Earth's resources, but also to solve problems of optimizing the composition of the soil for different climatic zones or different forms of agriculture. Finally, multilayer images can include superimpo sed maps of roads and drainage systems, and city communciations and ne tworks. It is important to note that a multilayer (or multizonal) anal ysis of all the above images is best carried out when and only when th ere are correlations between the layers and it is necessary to investi gate or estimate their mutual influence. Although the problems involve d in forming images in separate spectral bands and the effects on them of atmospheric distortions, cloud cover, illumination etc. are known and are described in [1], the problem of an overall analysis of interz onal transitions without being tied to the physical nature of the subj ects has not so far been solved. In this paper we propose to analyze i nterzonal transitions using multidimensional-matrix structures. These enable us to introduce a number of new ideas which are easy to investi gate in practice.