MODERN METHODS OF REMOTE DATA PROCESSING

Abstract

Remote sensing is the modern technique of obtaining information about the objects by data collected from special apparatus that are not physically in contact with the objects being studied. Remote sensing provides information about objects at the surface of the Earth based on analysis of electromagnetic radiation reflected or emitted from these objects. Modern methods and technologies of remote sensing of the Earth make it possible to distinguish different objects at the surface of the Earth, to analyze these objects and changes, make predictive assessments. This technique is applied for numerous fields, such as: geology, ecology, geophysics, geography, hydrology, oceanography, meteorology. Remote-sensing application involves two distinct processes: data acquisition and data processing. One of the most difficult data processing procedures is the classification procedure. Classification is based on that fact, that different objects have a different spectral reflectance. Classification procedure is defined as the process of categorizing all pixels in an image to obtain land cover map. There are different classification approaches that have been developed and widely applied for solution of various tasks, such as maximum likelihood method, minimum distance method, parallelepiped classification, Support vector machines (SVM). It were analyzed main advantages and disadvantages of these methods. It also was noted, that many classification approaches can not process vague and conflicting information. It was proposed to apply the Dempster-Shafer evidence theory and Yager’s combination rule for classification of ambiguous and contradictory data from different spectral bands. The application of Dempster-Shafer evidence theory arises from the necessity to overcome some limitations of Probability Theory. It was also considered a formula of Yager’s combination rule in this work. It was noted, that Yager’s combination rule can quickly and easy process information. It was considered the numerical example, where Dempster-Shafer evidence theory and Yager’s combination rule were applied for classification in this paper. It was noted, that choice of classification method depends of specific task. Described methods for aerospace image processing can be applied in different actual ecological, agriculture and practical tasks.