Rubel A. Methods of service-oriented processing of remote sensing multichannel images

In the dissertation the actual scientific and applied task of developing service-oriented methods of remote sensing image processing, which take into account noise properties present in images, to assess the quality of original (primary) remote sensing images and potential filtering efficiency for decision-making on image denoising expedience has been solved. For the first time, the method for predicting measures (image quality metrics) of original remote sensing image quality using multilayer neural networks has been proposed, which, unlike the known ones, uses image statistical parameters in the spatial and spectral domains in the absence of reference (noise-free) image and possibly without preliminary estimated noise characteristics. The method of predicting denoising efficiency of multichannel remote sensing images based on multilayer neural networks has been improved due to the use of the proposed set of input parameters (features), which allowed to increase the prediction accuracy in comparison with previously developed methods. The peculiarity of the proposed method is that it does not require any a priori or pre-estimated noise characteristics in contrast to the previously developed predicting methods. It is shown that prediction accuracy of the proposed method has been improved in terms of mean square error by 2-3 times compared to analogues, which facilitates more reliable decision-making on denoising expedience. The method of locally-adaptive filtering for radar images distorted by speckle noise has been improved using wavelet-based tetrolet transform, which allows increasing the denoising efficiency and at the same time to preserve edges, small-sized objects and textures. The results of the work have been used in research study at the O. Ya. Usikov Institute for Radio Physics and Electronics of the National Academy of Sciences of Ukraine and in the National aerospace university named after M. E. Zhukovsky “Kharkiv Aviation Institute”.