Research project

Operational concepts are getting more and more important. There are many possibilities to control interfaces, e.g. by speech or gestures. Each of these possibilities has its advantages and disadvantages. In this project we analyze how precisely we can determine the position of the hand with a PMD camera (Photonic Mixer Device) and the gesture class. The camera’s principle of operation is as follows: modulated infrared light signals are emitted by a sender on the camera, illuminate a scene, hit objects and are reflected back onto the sensor. Using the measured phase delay the camera computes the distances to the various objects in the observed scene. The camera also provides a conventional gray scale image. Both images are used for the segmentation of hand gestures.

The experimental setup is fixed and for the evaluation certain interfaces of the automotive environment are incorporated. The processing chain should take place in real-time i.e. one frame should be processed within 66 ms. The system is evaluated with a fixed test procedure to simulate near and far neighborhood relations.