Shape recognition in point clouds

OData support
Supervisor:
Dr. Csorba Kristóf
Department of Automation and Applied Informatics

In modern technology virtual reality, 3D visualization are getting more popular, that’s one of the reasons why point clouds are getting popular as well. Therefore it’s important that complicated point clouds are processed automatically and identification of certain shapes can be done precisely. In my thesis I analyze the available algorithms and methods to find objects automatically in a point cloud. To process the point clouds I used the open source Point Cloud Library, which gave me a wide range of possibilities to manipulate and change the point clouds. The goal of my thesis is to examine which are the most suitable classes and methods to recognize objects in a point cloud precisely using the Point Cloud Library. I examined in details the classes which implement the Random Sample Consensus, the Iterative Closest Point algorithm and Hugh transformation. During my analyzes I had to create basic programs to open a point cloud from an external file then implement the algorithms. In the following chapters I will analyze how these algorithms work, how to implement them and how precise is the given result.

Downloads

Please sign in to download the files of this thesis.