OrthoSwarm: Orthoimagery Drone Swarms
Tuhao Zhao, Peng Yi, Haozhou Zhai, Tianjiang Hu
AI summary
Problem
Existing orthoimage synthesis methods struggle with camera calibration requirements, high computational overhead, and memory limits, making rapid large-area mapping impractical for post-disaster drone swarms.
Approach
The system bypasses camera calibration by aligning drone imagery with pre-disaster satellite maps and rectilinear flight paths, using a parallel coarse-to-fine refinement pipeline to correct geolocation and scale errors efficiently.
Key results
- Calibration-free orthoimage generation leveraging pre-disaster satellite basemaps
- Parallel coarse-to-fine refinement architecture enabling scalable swarm processing
- New disaster-prone city dataset covering debris, waterlogging, and haze scenarios across 6.35 km²
- 6.35 km² orthoimage synthesized in ~11 minutes with quality matching commercial software
Why it matters
Provides first responders with a scalable, rapid mapping solution that overcomes calibration bottlenecks to accelerate post-disaster situational awareness.
Abstract
This paper addresses the urgent need for rapid synthesis of georeferenced orthoimages in post-disaster sce- narios, where pre-disaster satellite maps cannot be directly reused due to significant urban changes. Drone swarms offer advantages of large scale, wide aerial view and rapid coverage, of disaster-stricken areas. However, synthesizing georeferenced orthoimages within limited time remains challenging without camera calibration, primarily due to inevitable inconsistencies in intrinsics and extrinsics across different cameras, as well as sensor errors. To tackle this issue, we propose OrthoSwarm, a parallelizable calibration-free system architecture that leverages drone swarms rectilinear path planning and pre-disaster satel- lite maps for efficient orthoimage synthesis. OrthoSwarm’s per- formance is validated on a self-constructed benchmark dataset, generated by drone swarms in a digital twin city covering 3 natural disaster scenarios(debris, waterlogging, haze), with real- world validation using real single-drone aerial videos split into segments to simulate swarm acquisition. Experimental results from both simulated and real-captured data confirm the effec- tiveness of the proposed approach, enabling fast and visually consistent georeferenced orthoimage synthesis in stable post- disaster environments to support first responders promptly.