Abstract: Long-range vision-based measurement with unmanned aerial vehicles is highly susceptible to geometric degeneracy and sparse feature distribution, which can significantly degrade measurement accuracy in target localization and environmental modeling. To overcome these limitations, a landmark layout optimization method was proposed. By incorporating long-range imaging geometry and observation noise modeling, and by jointly considering average reprojection error, Fisher information matrix and spatial distribution as evaluation metrics, both measurement accuracy and observability were comprehensively assessed. Simulation results indicated that the reprojection error decreased by approximately 66%, the stability of pose estimation was enhanced. Specifically, the accuracy of visual measurement improved from 60.87 m to 0.79 m. These findings demonstrated that the proposed method effectively mitigates geometric degeneracy in long-range scenarios, enhances measurement robustness, and provides a transferable optimization framework for high-precision long-range measurement tasks with unmanned aerial vehicles.