Kalinina N. Mathematical models of the identification and methods of structural analysis of photogrammetric images

Thesis is devoted to developing models and methods for automated application of structural analysis and identification of color photogrammetric images. Proposed and justified the use of two-level structural segmentation algorithm for color images metals and alloys in the research of optical metallography, taking into account photometric invariance of images. A new mathematical model representation of color digital image finite set of evidence-based information semiinvariant characteristics. Proposed new information characteristics of digital color images, taking into account the physical mechanisms of fixation. A computational and methodological bases of identification of color images in the developed information and mathematical models for problems of optical metallography and recognition of space images of high spatial resolution. Based on the proposed mathematical model developed algorithms and application software authentication images. The basis of these methods laid apply information-theoretical approach for image projection display space information signs. Practical results form the prerequisites for solving applied problems of identification and interpretation of digital images of microstructures of metals and alloys, and data from remote sensing systems. The practical significance of the results obtained in the defined significant increase accuracy of identification of digital color images, which is the base of algorithmic mathematical model representation. The results of the research put into practice in the learning process.