A Novel Solution for Drone Photogrammetry with Low-overlap Aerial Images using Monocular Depth Estimation

TL;DR

Low-overlap aerial imagery challenges traditional photogrammetry. The authors propose a monocular-depth estimation workflow that uses tie-point–based metric-depth recovery to convert scale-ambiguous monocular depths into metric depths, enabling dense depth maps and complete scene reconstructions. Experiments on a Chongqing dataset show meter-level DSM accuracy and significantly improved completeness in non-overlapping regions, with Depth Anything v2 outperforming MiDaS v3.1 in edge detail and depth fit. The approach yields dense 3D products (dense point clouds, DSMs, orthomosaics) from sparse imagery, offering practical value for rapid drone surveys under time- and flight-constrained conditions.

Abstract

Low-overlap aerial imagery poses significant challenges to traditional photogrammetric methods, which rely heavily on high image overlap to produce accurate and complete mapping products. In this study, we propose a novel workflow based on monocular depth estimation to address the limitations of conventional techniques. Our method leverages tie points obtained from aerial triangulation to establish a relationship between monocular depth and metric depth, thus transforming the original depth map into a metric depth map, enabling the generation of dense depth information and the comprehensive reconstruction of the scene. For the experiments, a high-overlap drone dataset containing 296 images is processed using Metashape to generate depth maps and DSMs as ground truth. Subsequently, we create a low-overlap dataset by selecting 20 images for experimental evaluation. Results demonstrate that while the recovered depth maps and resulting DSMs achieve meter-level accuracy, they provide significantly better completeness compared to traditional methods, particularly in regions covered by single images. This study showcases the potential of monocular depth estimation in low-overlap aerial photogrammetry.

Figures (7)

• Figure 1: Low altitude, high speed, and significant ground variation can all lead to low aerial image overlap.
• Figure 2: The aerial photogrammetry solution suitable for low-overlap images.
• Figure 3: Data overview. (a) Black dots represent photo points from the original dataset, with red circles marking the selected low-overlap images after downsampling. (b) Distribution and overlap pattern of the downsampled images; the black rectangle indicates the validation area, measuring 400×400 meters.
• Figure 4: Depth recovery results for depth maps generated by different monocular depth estimation models. (a) The input aerial RGB image and the corresponding depth map generated by Metashape. Yellow points on the RGB image indicate tie points derived from aerial triangulation. (b) Results from MiDaS v3.1. (c) Results from Depth Anything v2.
• Figure 5: Additional Results of Depth Anything v2.