Multi-frame ship detection and tracking in an infrared image sequence

Abstract

This paper presents a multi-frame algorithm for detecting and tracking a moving target in the sea. We first analyse the properties of the images by splitting the images into several horizontal slices and then restoring the distortion area by lowering its gray levels. For target detection, the candidate target regions are extracted by a simple static thresholding method, which is, however, dependent on the strength of reflection. The regions extracted may not correspond to the actual shape of the real target. If some regions exist for several frames, they are taken as the regions of the real target. The system then enters the tracking stage. By moving the centroid of the current tracking window in the tracking window of the previous frame, the contrast-difference correlation is calculated to determine whether the system remains in the tracking mode. If the tracking stage remains in the coasting mode for several frames, the tracking fails and the system goes back to the detection stage. Experimental results demonstrate the applicability of the proposed system.