Object rotation axis from shading

In this paper, a non-complex estimator is developed for the tilt angle of the rotation axis of an object that is illuminated by a point light source. The tilt angle defines the orientation of the 2D projection of the 3D object rotation axis in the image plane and is a strong clue for image understanding. The estimator evaluates two images of a video image sequence showing a moving object. Additionally, only a displacement vector field and the 2D object silhouette are required. No 3D information is required. Therefore, the object is assumed to be rigid, to be matte, to have equally distributed surface normals and to be illuminated by a distant point light source and ambient light. For estimation, the displaced frame ratio (DFR), i.e. the frame ratio after motion compensation, is evaluated statistically. The DFR depends only on the photometric effect of temporally changing object shading. Experimental results with real images show the proper performance of the derived estimator for real objects