Enhanced feature detection and matching under extreme illumination conditions with a HDR imaging sensor
Lounis Chermak, Nabil Aouf · 2012
The performance of vision based autonomous navigation systems relies heavily on the characteristics of the cameras used to collect the data from the environment. This is even truer for environments subject to challenging illumination such as that experimented in space missions. This work investigates the level of performance that can be achieved with the use of a high dynamic range (HDR) imaging sensor compared to a color HD digital camera. The specific interest of this work focuses on the evaluation of feature detection and matching operations within different scenes presenting extreme illumination conditions including indoor and outdoor environments subject to direct sunlight exposure and also dark and low light indoor environments. This study examines within these distinct lighting conditions five different techniques for feature detection and two feature descriptor methods for matching. The results show a marked increase in the matching performances for images taken with the HDR imaging sensor from around 2 to 30 times better on average compared to the HD digital camera.