Efficient lookup table based camera pose estimation for augmented reality

Until now, existing camera pose estimation methods for the widely used square marker-based augmented reality (AR) are either highly sensitive to noise or much time consuming, and developers have to work hard to find the proper trade-off between computational speed and quality in mobile AR applications where computational resources are limited. The major difficulty is that only the 4 corner points of the square AR marker are available, and no redundant point correspondences can be used for a stable estimation. To solve this problem, an efficient lookup table (LUT)-based non-iterative solution is presented in this paper that achieves high stability in the presence of noise better than the most robust and accurate iterative solutions in the field, with the same level of accuracy and a much lower computational complexity. Our central idea consists of extracting a key parameter β from the camera pose and creating a LUT for β by