Metody shlukování ve zpracování obrazu

Abstract

This bachelor thesis is focused on informing about digital image processing with lay emphasis on clustering methods. I inserted clustering methods to larger complex segmentation and specify how it serves, how we can divide it and which basic segmentation methods are known. I got down to clustering methods in another part. I divided them also described their functions and principles. I choose clustering method mean-shift for example. Implementation was made by support of library OpenCV in programming language C++. It followed set of tests which gave us enough information about algorithm functionality and his properties. Mean-shift segments even more difficult diagrams (doesn´t has any problem with long clusters) also there is one more advantage to set final quantity of clusters it is not necessary.