A localization algorithm is developed to assist automated landing on unknown planetary surface. Classically, using a vision-sensor only, the vehicle states are subject to an observability issue. In order to overcome this problem, relative motion estimates were used as measurements in addition to image-plane data of the feature points. Using these data as measurements, a nonlinear least square estimator was designed that estimates the state vector when a priori knowledge of the state is not available. Furthermore, an Extended Kalman filter with fusing on-board IMU data was developed and shows promising results for future refinement of the previous estimates.