Research object - the processing and classification of objects on images.Research target is the development of models and methods of intellectual expedited grayscale images processing that can analyze information about the size of the image and load flow for a uniform distribution of the adaptive tasks to threads. Methods of research: for the task used in set-theoretic approach to the creation of a generalized process model of halftone images, methods, image processing, digital signal processing means, linear algebra and matrix theory, the concept of high-computing, computational methods for the development of accelerated processing models of visual images; graph theory to create models of image processing; simulation, which confirmed the effectiveness of the results; artificial neural networks of perceptron type. Scientific novelty: 1. The tier-parallel models of line and block halftone images processing, which due to the preliminary allocation of the source image into groups of rows or blocks avoid the "curse of dimensionality" and accelerate the process of evaluating images are first proposed. 2. For the first time a method of accelerated binary image sceletonization based on bit patterns, which due to the simultaneous use of thirty six bit patterns to accelerate the process of allocating the center line of the image, as well as to reduce the computational complexity in contrast to existing iterative sceletonization's methods is proposed. 3. Generalized model of the processing of halftone images, which, unlike the existing ones, provides parallel execution of binarization, neural network classification, which can significantly reduce the processing time of input halftone image is improved. 4. Method for the classification of halftone images using an artificial neural network perceptron type is improved. Unlike the existing ones, improved method automatically adaptive uniform distribute groups of neurons in the flows, depending on image size and download streams for parallel computing to accelerate the image processing of high dimension. The degree of implementation - a distributed system for determining the quality of bricks at its exit from a tunnel dryer was developed and used in a limited liability company "Amper" (Kharkiv, Ukraine) (the act from 23.12.2009, the introduction), a parallel system of automatic competitions' judging was developed and implemented in the junior sports school № 1 (Kharkiv, Ukraine) (the act from 20.01.2009.), the results of the thesis was also introduced in the learning process at the Department of Electronic Computers in the disciplines of "Intelligent Computer Systems", "Parallel and Distributed Computing" and "Methods and Means of Computational Intelligence" of Kharkiv National University of Radioelectronics (the act from 10.09.2010). The scope of use - for processing and classification systems of large amounts of input data, which can be obtained from different sources (photo-or video cameras, thermal imagers) that are based on parallel-series models, according to the concept of parallel and distributed computing and high-computing systems, intelligent systems in the classification all sorts of information in the learning process during training in the field of parallel computing technologies in image processing tasks.
