Markless tracking based on natural feature for Augmented Reality

This paper presents a solution for AR based product information displaying with virtual 3D model inserted into real world scene. In particular we developed an AR system based on markless tracking which can deal with different illumination conditions during the tracking stage, partial occlusions and wide view angle. Through FAST algorithm the AR system can real-time detect corner on image. In order to confirm the relative position of the camera and the image L-K optical flow algorithm is used to track the feature points in this AR system. The algorithm of K-D tree and RANSAC are also used to help the system to determine the camera parameter. Experimental results show the efficiency, accuracy and robustness of this proposed approach.