Abstract: A vision guidance scheme is proposed to actualize the autonomous returning and landing of a novel unmanned helicopter(NUH),and an extended Kalman filter(EKF) algorithm fusing inertial data and vision data is designed for the vision guidance scheme.The inertial data are used for recursive estimation of the NUH position.The vision data,i.e., the central point position and the size of the guidance target image acquired after image processing,are used to revise the accumulated errors of the inertial recursive estimation.The more accurate achieved NUH position estimation is used as the position feedback signal necessary for NUH vision-guided control.The position estimation effect and time cost of EKF algorithm are acquired through EKF simulating calculation.The feasibility of the proposed vision guidance scheme is demonstrated with the vision-in-loop guidance and control simulation.