Measuring Behavior 2005: Andersen and Due Hansen

Measuring the image quality of a car based sight effectiveness enhancement system

H.H.K. Andersen and K. Due Hansen

System Analysis Department, Risø National Laboratory, Roskilde, Denmark

Every year, numerous accidents happen on European roads due to bad visibility (fog, night, heavy rain).

Infrared images have the potential of augmenting the out-of-the-window view in cars at times when environmental conditions (e.g. darkness and fog) hinder optimal visibility and several studies have supported their beneficial effects.

In the present study we assessed the perceptual qualities of infrared images. Our experiment was based on one CCD video recording and three types of infrared recordings: long wave, short wave and data fusion of the two. The videos were mixed using a quad mixer giving us four synchronously running displays in one recording. The videos were recorded during nighttime driving.

The purpose of the study was to evaluate which of the four available displays the participants prefer under varying conditions. The primary measurement was eye movements recorded with a remote eye-tracker. The experiment was conducted as a within-subject design, involving 8 participants. Participants were asked to look at 16 video recordings. We used detection (detection of high emissive objects e.g. humans) and maneuver tasks (low emissive terrain features e.g. road signs) for measuring the participants’ preferences. Data includes gaze time on areas of interest and fixation frequencies. Each display was defined as an area of interest. A fixation was calculated every time the eye was fixed for a minimum of 120 ms.

The results of the analysis of the eye-tracking data show that that the participants in detection tasks looked at the long wave infrared display for significantly longer periods of time compared to the other displays. The results from the maneuver tasks show that the data fusion performs at the same level as the long wave. For the system to be implemented in a car these results tend to indicate that the driver should be offered the possibility to have both display types available at the same time or at least to choose between them according to his preferences.

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.