Distributed computing system for preprocessing of multichannel and multirange images

The paper proposes an approach to the formation of systems with distributed calculation of operations and presents the structure of the formation of the primary processing block of both image frame series and multi-range data. The presented structure is based on the possibility of parallel data processing. A multi-criteria filtering algorithm was used as an algorithm for improving data quality. The proposed algorithm allows for simultaneous minimization of a function with the criterion of the mean square difference between estimates and input implementation and the criterion of the mean square deviation of neighboring elements. The ranges of change of the control parameters of the proposed algorithm for dividing a twodimensional signal into objects are presented: stationary sections; boundaries; complexly textured blocks. The calculation and comparison of the speed of execution of operations using distributed calculations are presented. A comparison of the original and proposed approaches for processing a series of images and a pair of related frames obtained in different electromagnetic ranges is given. Pairs of test images obtained by sensors fixed at different magnifications with a resolution of 1024x768 (8 bits, color image), IR spectrum 320x240 (8 bits, color image) are used as test data used to evaluate the efficiency. Images of simple shapes are used as analyzed objects.