Research on image motion vector processing in electronic image stabilization

Image motion vectors are gotten by interframe matching and are filtered to be separated into intentional motion vector and image dithering noise which caused by instability of optical system's carrier. Actual compensating vectors are gotten. The multi-vector averaging strategy was presented and practically ameliorated to achieve better response to intended motions, so image stabilization can be attained even if the camera is being panned and intentional camera motions are not noticeably affected. By simulation on software programming and experiments on real hardware electronic image processing system, the method was proved to be simple, reliable and it performs well on condition that the motion vectors are logically calculated.