The paper addresses the control of Vertical Take Off and Landing (VTOL) Underactuated Autonomous Vehicles (UAVs) in hover flight, based on measurements provided by an on-board video camera and rate gyros. The objective is to stabilize the vehicle to the pose associated with a visual image of a planar target. By using the homography matrix computed from the camera measurements of the target, stabilizing feedback laws are derived. Explicit stability conditions on the control parameters are provided. It is shown that very good robustness and performance can be achieved without any apriori information on the visual target (like geometry, or orientation), by a proper tuning of the control parameters. Simulation results confirm the effectiveness of the approach.