The aircraft landing on carrier remains a very challenging task whose success affects the efficiency of the entire aeronaval group. In the French Navy, aiding systems improve the pilot perception during the approach toward the carrier. However no automatic landing system currently exist. This kind of system will be required for use of UAV. This thesis develops an automatic landing system based on a vision sensor of the aircraft. Using aircraft sensors, various functions of such a system include the carrier detection in the initial image, its tracking along images and the control of the aircraft and camera orienta- tion. This study presents solutions for each function based on state of the art techniques. Ship detection relies on aircraft sensors and on a reference image to provide tracking initialisation. Carrier tracking is performed using 3D model-based or 2D dense trackers. Aircraft control by visual servoing uses visual features of the image plane to perform the alignement and descent tasks. Desired values of the visual features remain constant along the desired trajectory. Car- rier movements are taken into account using vision. Control of camera platform is also based on visual features. First, vision methods are validated both using real and synthetic images from a realistic simulator. The aircraft control is evaluated with simulated visual features in order to characterize it. Finally, the complete system based on vision and visual servoing functions is successfully validated on realistic simulations.
