A High-Fall Scene Reconstruction Method Based on Prominence Calculation

A novel method is proposed for reconstructing high-fall scenes in complex urban environments based on prominence maximization, addressing the limitations of traditional UAV tilt photogrammetry. Initially, UAV’s initial acquisition data is used to generate a Digital Surface Model (DSM). An algorithm is designed to identify local wave peaks and calculate their saliency values, which, combined with linear interpolation, results in a smoother surface elevation model(SDSM). This provides new planning points for the camera’s position, overcoming issues like mismatching shadows in the DSM model. The method was tested in Boston, USA, covering various functional types and traffic scenarios. Results indicate that compared to traditional tilt photogrammetry, UAV tilt photogrammetry based on saliency maximization achieves higher accuracy, effectively capturing local details in the surveyed area while ensuring flight safety. Furthermore, the actual 3D model’s texture information and geometric structure are expressed with high quality.