Rubel O. Filtering and denoising efficiency prediction methods for spatially-correlated noise suppression in images in RS systems

The research object - processing of single and multichannel optical and radar RS images with spatially-correlated noise; the research goal - to improve the efficiency of spatially-correlated noise suppression in images obtained by RS systems and development of filtering effiiciency prediction methods; the research methods - methods of probability theory and math statistics, particularly, regression analysis, numerical modeling, theory of nonlinear and adaptive filtering, spectral correlation analysis; practical results - efficiency of spatially-correlated noise suppression has been improved for joint filtering of several channels in hyperspectral RS images, statistics of DCT coefficients has been used for operative prediction of denoising efficiency indexes, practical recommendations for filtering applying have been given; novelty - roubast metrics Canberra and Bray-Curtis have been used for nonlocal denoising methods for the first time, the method of multichannel RS images filtering has been improved, statistics of DCT coefficients in blocks has been used for the first time for denoising efficiency prediction, methods of filtering efficiency prediction have been improved; degree of implementation - results have been implemented in KhNUAF named by Ivan Kozhedub, CASRE IGS NAS of Ukraine; area of use - remote sensing.