GET-Forschungsseminar Abstracts

Thermal Visual Attention, Multi-Sensor Fusion

Dorid Kheirbek , GET Lab

Vortrag: Mi. 16. November 2011, 16:30, Raum P 1.4.17

Zusammenfassung:

Within a rescue scenario, the detection of victims has high priority. One task is to determine their vital functions. Therefor multiple sensors are required to cover the wide range of signals one might get from a human victim. The master thesis aims at integrating a thermal camera in a visual attention model, giving high priority to objects in certain temperature range and visualizing them in the video camera frame. This task is tricky due to the fact that the sensors should operate on a moving robot and the two cameras have different fields of view, different resolutions, and different optical axes. In order to solve these problems several approaches are investigated, then tests and experiments were done to show which of these approaches has better potential to improve the victim detection process in a rescue robot system.