PROCESSING AND USAGE OF THE MULTISPECTRAL IMAGES IN AGROMONITORING

Abstract

The paper contains the analysis of the process of images processing in agromonitoring, that enables to obtain complex information, regarding the state of the fields and agricultural crops, cultivated on these fields. This implies the passage of several stages of the complex preprocessing, which performs the correction and improvement of the space survey and aerial survey images and subject-related processing, which provides the decoding process or recognition of the objects and phenomena on the space survey or aerial survey images. Each of these stages includes the corresponding types and ways of the realization, besides some methods of image improvement provide the change of the spectral characteristics of the image, that is why, after their application the methods of subject-related processing, based on the analysis of the values of spectral brightness of pixels (classifications, arithmetic transformation of the channels, etc.) can not be used. In the given research on the base of the analysis of the literature sources, types and ways of the multispectral images processing, regarding agromonitoring are distinguished, as the work with such images due to a small number of spectral bands requires less cost for the processing without quality degradation unlike hyperspectral images. Types of the preprocessing stage are considered, namely geometrical correction of the images, radiometric calibration of the images, radiometric correction of the atmospheric impact, restoration of the missed pixels, contrasting by means of the histogram linear spreading, contrasting by means of histogram equalization, filtration. Methods of the subject-related processing stage are considered, namely: color transformation, analysis of the main components, spectral division, classification with and without learning, image indexation, determination of the quantitative indices. Their peculiarities, conditions of usage, etc., are described. The procedure, describing the stages of multispectral images processing in agromonitoring, taking into account all the types and methods of each stage realization, is suggested.