Augmented reality for planar scene via affine transform based natural features tracking

This paper illustrated a novel registration method for planar scene based on natural features tracking via affine transform. Our method can be divided into two stages: offline 3D reconstruction and online registration. With the obtained feature correspondences, the homographic was computed between the world plane and the current frame robustly. Then, the registration matrix was synthesized based on the above homographic and the transform matrix between the plane and world coordinate system obtained in offline stage. The proposed method also distinguishes itself in two ways: Firstly, a new method was established in the world coordinate system. Secondly, a novel affine transform based method to match features between reference images and currently frame directly. The method can avoid the user's intervention in initializing stage, while overcoming the problem of error accumulation. Some experiments were carried out to demonstrate the validity of the proposed approach.