NONPARAMETRIC METHOD FOR DETERMINING THE NUMBER OF CLASSES IN AN IMAGE SEGMENTATION PROBLEM
R. V. Podrezov1, M. A. Raifel'd2
1AO "Nauchno-Issledovatel'skii Institut Elektronnykh Priborov", Novosibirsk, Russia
2Novosibirsk State Technical University, Novosibirsk, Russia
Keywords: сегментация изображений, гистограмма рангов, ортогонализация Грамма-Шмидта, метод главных компонент, image segmentation, rank histogram, Gram-Schmidt orthogonalization, principal component method
Abstract
In problems of automatic threshold image segmentation by brightness, the important questions are those about the number of brightness classes and, as a result, the required number of thresholds. A solution to the problem of estimating the number of classes in an image is often based on representing its distribution as a mixture of distributions of brightness classes. It is known that this problem (splitting the mixture) has a solution only for certain types of distributions, and, if the distributions of brightness classes are unknown, difficulties arise in its application. In this paper, a nonparametric method for determining the number of classes is presented, based on rank histograms and using the property of local spatial grouping of elements of each brightness class in an image. Comparing the proposed method with various criteria for estimating the number of classes in images shows that the method under consideration is effective
|
