Gaussovský "blurred mean shift": implementace a experimenty

Abstract

The aim of this bachelor thesis is to get familiarized with the method in image segmentation called Gaussian "Blurring Mean-shift", its effective implementation and then experimentally verify its behavior. In contrast to human, computer is unable to recognize objects in the image. This do segmentation algorithms. The introduction deals with the description of the method "mean-shift", as one of efficient algorithms for image segmentation. In the next section, I describe directly the Gaussian "Blurring mean-shift", its function and the possibility of its acceleration. Implementation was made with multiplatform library Qt, Qt SDK and programming language C/C++. I used some experiments to get familiarized with behaving of the methods and my results afterwards documented in this thesis.