Pedestrian detection with night vision systems enhanced by automatic warnings

Abstract

The experiment described in this paper is a follow-up to a prior experiment conducted by the authors on night vision systems. In that experiment, 2 major sensing technologies for automotive night vision systems were compared in a laboratory experiment: far infrared (FIR) systems which generate images by passively detecting thermal emissions from objects and surfaces in the road scene, and near infrared (NIR) systems which actively illuminate the scene in the near infrared spectrum and capture the reflected radiation. It was argued that pedestrian detection should be the main objective of any night vision system. The authors compared detection distances for pedestrians in both systems using matched stimuli and found that detection distances with FIR were overwhelmingly greater than with NIR. The experiment described in this article extends the prior experiment in 2 ways. First, subjects viewed the night vision display intermittently as they had to perform a concurrent simulated driving task that required viewing the forward scene almost continuously. Consequently, their overall performance was expected to degrade, but it was not clear by how much and whether the degradation would be proportional for both night vision systems. Second, some of the night vision scenes were artificially enhanced by automatic pedestrian detection at 2 set distances from the pedestrian. As discussed in this article, automatic pedestrian warnings are receiving great development interest and are beginning to become available on new vehicles. It was expected that pedestrian warnings would improve performance but it was not clear how they would change the drivers' scanning pattern of the display and whether the distance at which warnings appear would affect detection distance and detection probability.