| Abstract [eng] |
In the dissertation, methods for processing of digital images are provided. Software programs for identification of digital images upon applying the theory of covariance functions are presented. An analysis of numerical correlation of digital images was performed on the base of the theory of covariance functions in different quantization intervals. Identification of changes of the Earth’s surface caused by various factors (such as human activities, erosion and so on) in a locality takes much time. Supplemental devices are required for fixing and assessing such changes of the Earth’s surface. In the present work, an opportunity to identify a similarity (dependence) of digital images in an automated way is provided. An interdependence of the images shall be assessed according to the changes of the values of correlation coefficient in the range of RGB colours. The dissertation consists of introduction, four chapters and the general conclusions. In the Introduction, the problem under research is substantiated, the relevance of the problem is defined, the objective and the tasks of the work are formulated, the scientific novelty is described, the statements of the dissertation under defence are provided, publications and conference reports of the Author are presented, and the structure of the dissertation is described. Chapter one, an analysis of scientific references is provided. In it, the existing clustering methods and their application are discussed upon. The structure of digital raster images, the existing GIS methods and their application in analysis of digital images are reviewed. The tasks to be accomplished in the dissertation are formulated. Chapter two of the dissertation, the theoretical research is provided. The theoretical presumptions of correlation methods and the issues related to application of covariance functions for analysis of digital images are discussed upon. Chapter three of the dissertation, the research works on identification of digital raster images are presented, the impact of different ranges of the spectrum of the digital images upon the results of computation is assessed. The peculiarities of computation on changes of values of parameters of digital images are established. Chapter four of the dissertation, an application of clustering correlation analysis in automated land cadastre accounting procedures is provided. Four papers on the subject of the dissertation were published: three of them were published in journals referred in international databases and one – in the reviewed materials of international conference; in addition, 6 reports were delivered at conferences. |
