(An) object tracking method using FLIR image of approaching ground target

Abstract

It is hard to track targets in images acquired from Forward Looking Infra-Red (FLIR) image tracker on fire-and-forget missile because the target in IR is rapidly changed according to distance and angle between the IR camera and target on ground. The target in image continuously grows from frame to frame and positions of target are not fixed due to large variation of camera on missile. Clutters near the target are also obstacles to track the target successfully, and the thermal signature of IR image effects on the target tracking especially in terminal homing stage. This thesis focuses on solving related problems and on improving of target tracking algorithm for FLIR image tracker on fire-and-forget missile. We use template matching method to find target in each frame. The new template is updated from previous frame around the object center which we searched with template matching. The size of continuously growing template is predicted by considering the trace of missile and characteristics of IR image seeker. To make well-centralized template, we project image gradients to vertical and horizontal direction and experimental results show the proposed method accomplish target tracking successfully.