Open AccessСЕГМЕНТАЦІЯ ДВОВИМІРНОГО СИГНАЛУ, ЯКИЙ ПРЕДСТАВЛЯЄ СОБОЮ ЗОБРАЖЕННЯ ТОВАРУ, ЩО ЗАМОВЛЯЄТЬСЯ СПОЖИВАЧЕМ
Author(s) -
Володимир Олександрович Патрушев,
Ольга Ігорівна Патрушева
Publication year - 2019
Publication title -
radìoelektronnì ì komp'ûternì sistemi
Language(s) - English
Resource type - Journals
eISSN - 2663-2012
pISSN - 1814-4225
DOI - 10.32620/reks.2019.1.04
Subject(s) - particle swarm optimization , metaheuristic , cluster analysis , simulated annealing , computer science , image segmentation , segmentation , artificial intelligence , preprocessor , segmentation based object categorization , scale space segmentation , multi swarm optimization , swarm behaviour , pattern recognition (psychology) , mathematical optimization , data mining , mathematics , algorithm
The subject of study in this article is the means of segmentation of the image of the ordered goods by the consumer online store. The goal is to determine the means of segmentation of the image of a two-dimensional signal. Objectives: analyze existing methods of image segmentation, select metaheuristic clustering with an interactive task of the number of clusters, conduct research. Methods used: segmentation of a two-dimensional signal, which is an image of a product ordered by a consumer in an online store. A meta-heuristic clustering method was implemented with an interactive assignment of the number of clusters. The method is based on the optimization of particle swarm (PSO) and annealing simulation (SA), an adaptive optimization of particle swarm (APSO), which underlies the image segmentation, is proposed. The following results were obtained. The use of simulated annealing in the proposed adaptive optimization of a particle swarm provides: control of the rate of convergence of a given metaheuristic method; research in the early stages of the entire search space, and in the final stages - the focus of the search. To determine the effectiveness of the proposed method, studies have been conducted that prove that the mean square error does not exceed 0.05, which in turn proves the effectiveness of the chosen method in image segmentation. Conclusions. The scientific novelty lies in the fact that to solve the problem of determining the method of image preprocessing, a clustering method with a given number of clusters was used, namely a metaheuristic method based on optimizing the particle swarm (PSO) and simulating annealing (SA) using adaptive particle swarm optimization (APSO), which underlies the image segmentation. The use of simulated annealing in the proposed adaptive optimization of a swarm of particles provides: control of the rate of convergence of a given meta-heuristic method and research in the early stages of the entire search space, and in the final stages the direction of the search. As a result of a numerical study, it was found that the mean square error does not exceed 0.05.