Estimation of rotation and divergence from optical flow constraint

In previous paper[1], we suggested a new algorithm to estimate optical flow and motion parameters (angular velocity and scale of enlargement). There are two advantages in this method. First, more accurate optical flow and motion parameters can be estimated because it is strictly consistent with the assumption that the motion contains not only translation but also rotation and divergence. Second, this method can reduce the cost of calculation because optical flow and motion parameters can be estimated at the same time after scanning images only once. We examined the effectiveness of the proposed method in some simulations in previous paper. In this paper, we will show experimental results of proposed method for more realistic scenes.