Ponomarenko M. Methods of assessment of visual quality and similarity of digital images for telecommunication tools

Українська версія

Thesis for the degree of Doctor of Science (DSc)

State registration number

0512U000625

Applicant for

Specialization

  • 05.12.13 - Радіотехнічні пристрої та засоби телекомунікацій

27-06-2012

Specialized Academic Board

Д 64.062.07

National Aerospace University "Kharkiv Aviation Institute"

Essay

The object of research - the processes of formation, analysis, processing and transmission of images, video sequences and one-dimensional data arrays that are used in telecommunications tools.; the aim of research - development of efficient methods of assessment of image visual quality and similarity for telecommunications, as well as their use in the design and performance analysis of digital image processing methods; methods of research - methods of statistical processing of spatial random processes and fields, methods of spectral-correlation analysis, methods of nonlinear and adaptive filtering, numerical optimization techniques and mathematical modeling, probability theory, the theory of encoding data, the theory of orthogonal transforms; results - efficiency of verification of visual image quality metrics is increased, efficiency of assessment of image visual quality is increased, efficiency of assessment of image similarity is increased, efficiency of blind estimation of the parameters of image distortions is increasing, compression ratio of images is increased; novelty - full reference and no-reference metrics of visual image quality are developed, multiparameter measure of visual similarity of images is developed, methods of database forming for verification of image quality metrics are improved, a method for fast clustering in multidimensional space based on the use of set boundary elements is developed, a method of simultaneous blind estimation of noise variance and spectrum is developed, visual quality measures for verification of image denoising methods are improved, method of testing for Gaussian data samples which correspond to a mixture of Gaussian and generalized Gaussian distributions is developed, method for automatic selection of filter parameters and image compression parameters based on the evaluation of visual quality is improved, a recursive method for data compression based on grouping of is proposed; the degree of implementation - implemented; area - instrumentation.

Files

Similar theses