Restoration of vibrated images can be performed with or without knowledge of the motion function. The algorithms based on the known motion function are more accurate. However, sensing of the motion function in the real-world environment is often difficult to implement accurately. We propose a differential scheme for sensing vibrations. It is shown that the differential scheme permits one to obtain a true rotational motion function, which is often the main blurring factor in remote sensing systems. An experiment was set up and the results are presented. The proposed sensing scheme was implemented in real time and then the obtained image was restored from motion blur with a personal computer. The fine accuracy of the restoration results derives from the accuracy of the proposed motion sensing scheme.