Akhmetshina L. Segmentation and quality increasing of low contrast images on base of local-adaptive transformations

The thesis is devoted to statement and solution a problem of low contrast images processing under conditions of a priori uncertainty system characteristic of theirs forming and also spectral and statis-tical properties of useful signal (region of interest) and a noise. Using local-adaptive transformations of brightness images gives possibility for transition at a new information basis. It opens a practical possibility for solution the task increasing quality, sensitivity and resolving power visual analysis of low contrast images, and also validity of segmentation procedure on base using virtual analogies with most high-accuracy and sensitive methods of radiophysical and optical measurements (holography, interferometry, ellipsometry) and corresponding of them mathematical apparatus. Results experimental testing of devised methods on digital models and real images are presented. A domain of ones applicability is identified and made comparison with known methods of image processing. Information possibilities methods of multiparameters neural network and fuzzy clusterization are developed and research. Means forming a new "composite" image on base of multidimensional synthesized ensembles various characteristics of analyzed images are presented. It gives possibility for facilitating multiparameter data analyzing and interpretation. . Experimental results confirm the efficiency of the proposed methods on a number of well-known benchmarks and the real fuzzy data classification problems.