Detecting Search and Rescue Targets in Moving Aerial Images Using Eye-Gaze

摘要

Eye gaze performance is measured and characterised in the context of moving image inspection. The application domain is that of Wilderness Search and Rescue. Here, moving terrain images captured by an Unmanned Aerial Vehicle (UAV) must be inspected accurately and with minimal delay to attempt identification of a missing person or related artefacts. An exploratory pilot experiment using images of real terrain established the extent of gaze for North and East movement of the UAV at four different speeds, and also for the condition in which a sequence of static frames of the same moving image was inspected. Observation were also made of the gaze tracking behaviour initiated by a potential target. Suggestions are made about the way in which the observation and characterization of eye gaze may be harnessed to enhance the success of target detection in a search and rescue context.