Camera pose estimation from image is one of the key issues in computer vision. Most existing approaches need embed a visual mark in the object's texture for camera pose estimation. However, a visual mark spoils the unity of the image. This paper introduces a novel pose estimation method using the spatial frequency domain analysis. In this method, the high frequency marker that is hard to distinguish is embedded in the image for pose estimation. We observe the spatial frequency change of the high frequency marker and infer the posture change of camera. The result shows that this method is feasible and flexible.
展开▼
