Tracking of sea-surface targets in infrared videos using scale invariant feature transform

In this study, sea-surface targets are tracked using scale invariant feature transform (SIFT) in real and synthetic infrared (IR) videos. The effects of the parameters such as number of features, contrast threshold value, corner coefficient, and the threshold used in matching features on the tracking accuracy are investigated in detail and target tracking success is evaluated with respect to the ground truth target positions. The results obtained show the possibility of developing effective target detection and tracking algorithms using SIFT in IR videos containing sea-surface targets.