A novel registration method based on triangular pattern marker

This paper presents a three-dimensional (3D) registration method based on a triangular pattern marker in augmented reality (AR). The triangular pattern on the marker in the captured video image is recognized via the edge detection. The triangle vertices which are used to calculate the homography matrix are extracted. The camera calibration for tracking accurate pose of the object is then implemented with the homography matrix that deduces into the camera extrinsic parameters. Through experimental testing with the method, the 3D virtual objects can be rendered accurately on top of the marker into the physical video image.