The rising demand for autonomous in orbit servicing (IOS) missions, ranging from inspection and refueling to complex docking maneuvers, is pushing vision based navigation to the forefront of modern Guidance, Navigation and Control (GNC) systems. Among these tasks, rendezvous operations remain particularly challenging, as they require navigation solutions that remain reliable over a wide span of conditions, precisely from long distance target acquisition to the sub-meter accuracy needed for docking. However, traditional GNC architectures often fall short in this respect, as they are not inherently designed to adapt seamlessly across such different operating regimes. To address these challenges, a dual-mode vision based navigation framework is proposed, tailored for fully autonomous rendezvous and docking in Geostationary Transfer Orbits (GTO). In particular, in the far range phase, the system employs a dynamic switching logic that alternates between pure Line-of-Sight (LoS) measurements and pixel based range estimation, both processed through an Extended Kalman Filter (EKF) to ensure stable relative state estimation. Once the chaser approaches the last 15 meters, the navigation shifts to ArUco marker detection performed on the docking face of the target. Thus, the camera pose is estimated via a Perspective-3-Point (P3P) algorithm and then refined with the Unscented Quaternion Estimator (USQUE), while a dedicated camera switching mechanism guarantees continuous marker visibility during the terminal approach. The proposed strategy was validated within the Horizon Europe GEORyder project, confirming that the combination of adaptive far range estimation and marker based close range navigation enables uninterrupted and accurate state tracking from approximately 1 km down to docking. By blending complementary filtering strategies with dynamic measurement models, the system achieves robustness, precision, and computational efficiency. These characteristics make it a promising solution for upcoming IOS missions, where flexibility, scalability, and reusability will be key drivers for success.