Abstract
In this research, we focused on the saliency map that predicts the visual attention based on analyzing the image of driver view, and we apply the saliency map as an index of the risk level of traffic environment. We proposed an evaluation method of intersection visibility using Saliency map, and investigated correlation between driver's timing of discovering intersection and visibility of intersection. From the experiment result, we confirmed that the coefficient of correlation is 0.88, thus strong correlation between them was confirmed. In addition, we proposed an alarm system according to the intersection visibility and evaluated the validity of the alarm system in a driving simulator experiment. From the experiment result, we clarified that brake start timing was delayed and sudden braking was often occurred at the low visibility intersection. Furthermore, we confirmed the effect of alarm system that induced a driver to discover the presence of the stop sign and operated the braking safely.
