We present a single-image algorithm for reconstructing the 3D velocity, the 3D spin axis, and the angular speed of a moving
ball. Peculiarity of the proposed algorithm is that this reconstruction is achieved by accurately analyzing the blur produced
by the ball motion during the exposure. We combine image analysis techniques in order to obtain 3D estimates, that are then
integrated into a geometrical model for recovering the 3D motion.
The algorithm is validated with experiments on both synthetic and camera images. In a broader scenario, we exploit this specic
problem for discussing motivations, advantages, and limitations of reconstructing 3D motion from motion blur.