Tracking of moving object using continuously adaptive mean shift algorithm with quadrotor unmanned air vehicle

Object tracking is the process of generating the trajectory of an object over time by discovering its exact position in every consecutive video frames. It is a challenging problem due to tracked object motion and size changes, illumination changes and egomotion. In this thesis, an approach that is able to cope with those problems using unmanned air vehicle was applied. Unmanned air vehicles has been an increasing field of research in both civilian and military applications. Among its different models, quadrotors have advantages such as high maneuverability and moving in three directions. The unmanned air vehicle used in this thesis is AR.Drone. The application is implemented using version 2.4.9 of OpenCV library with C programming language in Visual Studio 2010 environment. The application is capable of tracking moving objects by using the front camera of AR.Drone with resolution of 640x360 and at a frame rate of 30 fps. Tracking moving object is carried out in two stages. Firstly, Continuously Adaptive Mean Shift Algorithm on a static camera is applied to track the moving object and secondly AR.Drone front camera was used to carry out tracking a moving object with the same algorithm in real-time. Object tracking process is carried on independently from the distance between AR.Drone and the tracked object. In this thesis, object tracking is carried out succesfully with AR.Drone which is a mobile platform despite of changes in object‟s size and illumination.