Lau G. The Development of Antijamming Methods of Color Identification during Image Processing in embedded recognition systems

Dissertation is devoted to the development of new antijamming and invariant to the terms of illumination, hindrances and errors of receiver, mathematical methods and algorithms of color authentication for the image processing in object-oriented embedded recognition systems. Analytical dependences are offered for determination of influencing of illumination and receiver's distortions on the image with the purpose of choice of informative criterion which minimized such influence. The new method for determination of image pixels, which belongs to the informative fragment and which is described simultaneously in the image plane and in space color is developed. On the basis of the developed analytical model and implementation of competition mechanism, for the case when pixel belongs to few informative fragments, the row of algorithms of the coloured pixels filtration (selections) on the image surface is created. The program of quantitative estimation of areas of the coloured metallography samples, which realized the developed algorithms, is created. It is analysed the possibility of using developed algorithms in the embedded recognition systems on the base of MOEMS and appropriate architecture of the such microprocessor system is developed.