Měření vzdáleností mezi obrazy

Abstract

This thesis deals with issues related to measuring distance in images using chamfer matching. This measure is useful in detection of image templates in test images. The aim of this thesis is on methods, which are capable of detecting textureless objects. Only discriminative clue for these objects is their contour. Methods described in thesis can detect objects in scenes with cluttered background and in scenes with different illumination changes. There are four algorithms described in the text. Two methods are described comprehensively and they are implemented in practical part of thesis. Implemented algorithms are evaluated using common metrics and the results are compared to public implementations of algorithms presented in OpenCV library.