With the rapid development of commercial spaceflight and deep space exploration technologies, the guidance and control of spacecraft during soft landing has gained increasing attention from researchers and engineers. Traditional spacecraft guidance algorithms typically treat the translation and rotation of the center of mass as separate control objectives. However, in the actual rocket landing scenario, the thrust is the primary control input, resulting in a strong coupling between translational and rotational motions, a challenge that conventional methods cannot effectively address. Additionally, with the advancement of spacecraft technology, visual sensors, due to their high reliability, low cost, and adaptability, have become an increasingly important tool in space missions. In particular, in situations where satellite navigation systems are unavailable, visual sensors provide high-precision relative positioning data, greatly enhancing the reliability and safety of rocket autonomous navigation systems.
 
To address this issue, this paper proposes a vision-based servo navigation and guidance method using a pinhole imaging model. By using feature point information and distance data from image or laser sensors, and integrating the rocket's current linear velocity and angular velocity, a dynamics equation based on the image-based body frame is constructed. This approach effectively correlates the image information with the rocket's dynamic state, achieving coordinated control of both translational and rotational motions during landing, thereby improving landing accuracy and autonomy. Simulation results demonstrate that the proposed guidance strategy can effectively tackle control challenges in complex environments, providing a novel and effective approach for future spacecraft autonomous landing control.