At the on-orbit servicing final approach stage

the traditional monocular vision navigation method is limited by the camera field of view

and artificial feature markers cannot be placed on the non-cooperative target. To overcome the drawbacks

a vision navigation algorithm based on part of the solar array triangle support of non-cooperative spacecraft is proposed. Firstly

a "selfie stick" camera installation structure and a real-time camera calibration scheme are designed

and then the method of calculating the camera installation angle is provided. Next

the optimization model satisfying the real geometry structure of the triangle support is built based on the adverse projection theory. Then

the ant colony search algorithm is applied to solve the depth and the absolute position method is applied to calculate the relative attitude and position. Finally

the mathematical simulation in the background of on-orbit servicing final 2 m~0 m approaching of non-cooperative spacecraft is carried out. The simulation results show that the relative attitude and position determination accuracy are better than 0.2° and 3 mm when the relative distance is less than 1 m

and it proves that this algorithm is effective and feasible. The simulation results have proven the feasibility of the relative navigation scheme

and indicate that the proposed algorithm has high accuracy

which increases with decreasing relative distance between the servicing and the target spacecraft. Furthermore

the algorithm has good robustness to the measurement errors of projection points

and keeps the accuracy for relative large measurement errors.