Rozpoznávání objektů a jejich polohy na základě 3D modelu

Abstract

Thesis addresses object recognition and pose estimation using 3D object models, one of the key tasks in Computer Vision. Recognition itself is done in RGB-D image which was generated using Microsoft Kinect sensor. The goal of this thesis was to build a full recognition system using state-of-the-art algorithms included in PCL library. Then it was necessary to experimentally compare and evaluate used algorithms to see their overall successfullness with different parameters. Before all that a necessary theoretical basis is presented.