To address the limitations of traditional methods in complex spatial environments, an improved morphology algorithm based on light-dark threshold matching is proposed. This algorithm utilizes steps including image preprocessing, light-dark area identification, morphological processing, regional characteristic analysis, light-dark area matching, and circular fitting to achieve more accurate crater detection. Experimental results indicate that the algorithm achieves high accuracy and simplicity in crater identification, making it suitable for environments with limited computational resources. However, the accuracy of detecting craters with high elliptical eccentricity at the edges still needs improvement. Therefore, this paper also proposes an optimization method based on historical information sampling and parameter grid search to enhance the robustness and adaptability of the model.