An unscented Kalman filter (UKF) is derived for integrating vision with inertial measurements from gyros and accelerometers sensors based on three-view geometry. The main goal of the proposed method is to provide better estimations compared to the implicit extended Kalman filter introduced by Indelman [11]. The UKF uses a selected set of points to more accurately map the probability distribution of the measurement model than the linearization of the extended Kalman filter, leading to faster convergence from inaccurate initial conditions in estimation problems. The proposed method is validated using a statistical study based on simulated navigation and synthetic images data.