Development of an Alternative Text Input Device

OData support
Dr. Vajda Ferenc
Department of Control Engineering and Information Technology

Looking over the alternative text input devices of various school, I introduced in this thesis a hand gesture recognition system based on computer vision. First I dealt with a variety of alternative text input devices beginning with the touchscreens to the virtual keyboards and the analysis of brain and nerve impulses. Then I made a system plan for the solution, where in a gesture based environment the use of the keyboard can be avoided. My system detects and interprets the marks of the international dactile abc. In this thesis I introduced the hardware and software components of my system. Next I outlined the worked-out and applied algorithms. They comprise the segmentation, contour finding and processes for description and classification. The program follows the steps of a typical gesture-recogniton algorithm - segmentation, feature extraction and classification. I described the applied program packages too. In the first phase of the program development I worked under MATLAB environment. After successful implementation I used C++ language incorporating the open-source computer

vision and SVM software packages. In the thesis I made a system and a pilot application, where the user can type without keyboard. Finally I carried out experiments in order to determine the speed, robustness and running time of the program.


Please sign in to download the files of this thesis.