<title>Informational technology for investigation of structural dynamics of environmental objects by their multiple-spectral images</title>

Abstract

The informational technique intended for extraction of environmental objects upon multiple-spectral images and formation of characterictics describing in a sumarized way the dynamics of their spatial structure is being discussed. It is assumed that the multiple-spectral images make up a single vector field. The procedure of the object extraction starts from segmentation of random vector fields. The observation vector of the same space point leads to a qualitatively better process of marking of picture points belonging to different objects than one scalar observalion. An optimal algorithm is proposed based on vector observations and intended for segmentation of multiplespectral images.For extraction and classification of objects, a model is discussed of generalized structural spatial noise-immune description of multiple-spectral images. Basic principles of a system for the storage and fast retrieval of experimental and processed data are proposed. The results of experimenting with the technology in the domain of investigating natural processes by their satellite multiple-spectral images are presented.