Kobylin O. Methods and models of adaptive normalization systems of image processing

The dissertation work is devoted to developing an adaptive normalization mathematical model for computer vision systems. The existing methods of the images normalization are considered and their disadvantages are formulated. Fundamentally new approach to increase normalization methods efficiency lies in developing the new means of image data representation. For obtaining of delivered goal it is proposed to use the wavelet analysis which allows to analyze image and to determine geometrical features about an object. The method based on the mathematical model which allows defining a group of transformations is developed. The method of the allocation image edge using continuous wavelet transformation is proposed. The proposed criteria provide the correspondence to determine of the transformation groups, which allow to use their in real computer vision systems.