Yuzhakov S. Development and research applying of multiple transformations for high-performance computer image processing and image recognition

Dissertation deals with the improving of system performance of computer processing and image recognition by reducing the number of characteristic features and simplification of the computational process. To achieve this goal, it was proposed the use of parallel shift technology for recognition of not contour images. A mathematical model of the relationship between shape of the object and functions of the area of intersection was developed. Methods of image recognition, determining spatial orientation of objects and their movement parameters, determining the contour shapes were developed by analyzing a set of functions of the area of intersection. Software models for methods of image recognition and determining their movement parameters were developed to confirm the effectiveness of the proposed algorithms. The structure of the system and hardware to produce functions of the area of intersection of image and its copies that is parallel shifts were developed. The possibility to use the developed methods in the processing of both raster and non-raster images was substantiated. It was investigated reliability of the identification of images at different parameters of the input image by created program models.