Physical interfaces in augmented reality

OData support
Paróczi Zsombor
Department of Telecommunications and Media Informatics

Nowadays there is an increasing demand for new input devices, the focus of such developments are mostly creating fast, easy to use tools, but reducing the costs is getting more and more attention as well.

In my thesis I implemented an interface based on image processing, which provides an opportunity to fast and ergonomic communication without any expensive hardware.

A simple A4 sized sheet represents the communication interface between the computer and the user, on which there are different printed buttons. These buttons can be pushed virtually, by simple covering a button on the original picture. The implemented virtual input device is able to control multi media players with only one web camera.

In the first chapter I talk about the field and it's importance today. In the second chapter I describe more modern input devices, such as Kinect, Wii and GamePad with their benefits and disadvantages.

In the third chapter I present the structure of the implemented system and the initial state also. I talk about the algorithms that I used during the implementation and also describe the OpenCV library in detail.

During my work I used one of the most popular image processing method, in the field of the object recognition.

I present the training of the classifiers both in practice and theory, also the Haar features, the integral image and the cascade concepts.

I show how to create the dataset and I mention the Gaussian Pyramid and binary image methods too.

In the last chapter I summarize my work, furthermore I present the test results of the object detection and show how the recognition works on a practical example.


Please sign in to download the files of this thesis.