A variational approach for optical flow estimation in infra-red or thermal images

The most important application of the thermo vision system is the detection and tracking of moving objects in infrared images. In this paper, we have designed an approach for estimating the optical flow in thermal image sequences. This optical flow estimation approach is based on local as well as a global method which takes the advantages of both the brightness constancy assumption and spatiotemporal smoothness constraints. A new nonlinear cost function is designed with these constraints. This nonlinear cost function is minimized by the modified steepest descent method. The estimated optical flow fields demonstrate a high level of random motions caused by thermal noise. The experimental results show the robustness of the proposed method.