Robust centroid extraction using the hybrid genetic algorithm with applications to planetary optical navigation

Traditional radiometric tracking navigation increasingly fails to meet the demands of deep space exploration. In contrast, optical navigation enables interplanetary spacecraft to navigate autonomously with higher precision. The effectiveness of image processing algorithms plays a crucial role in determining the accuracy of optical navigation systems. This paper presents a robust centroid extraction method based on a hybrid genetic algorithm. First, noise interference is effectively reduced by leveraging proximity information. Second, a fitness evaluation mechanism is introduced to assess model performance throughout the iterative process. Third, an annealing mutation operator is incorporated to prevent premature convergence to local optima. Finally, extensive comparative testing demonstrates that the proposed method offers substantial improvements in both accuracy and robustness, thereby substantially improving the reliability of the navigation system under complex conditions.