UAV-based inspection is recently expected to enable cost-effective and accurate maintenance of civil infrastructures. This study aims to optimize UAV flight paths for inspecting steel plate girder bridges by prioritizing damage-prone areas as key viewpoints. Critical viewpoints are identified based on structural features and expected stress states under external loads, such as live or seismic loads. We construct a 3D model of a steel plate girder bridge using Autodesk Fusion 360 and Unreal Engine. Based on bridge inspection data analysis, critical viewpoints of interest (VPI) are identified and assigned importance levels. The flight path is then optimized to prioritize high-importance VPI while minimizing overall travel distance. Ant Colony Optimization (ACO) effectively generates efficient routes, ensuring both thorough inspection coverage and reduced flight time. Results show that optimized paths prioritize VPI near critical connections and structural elements, ensuring visibility and accuracy. The effectiveness of the optimized flight paths is validated through flight simulations in a virtual environment, confirming that the necessary viewing angles for inspection are achieved while minimizing travel distance.