SatOrtho

Satellite-Map-Prior-Assisted Real-Time Orthophoto Generation

Maoan Zhou · Yongfei Li · Min Kong · Jieyu Liu · Zhaoyun Luo · Dongfang Yang^*

Manuscript under review at IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (JSTARS)

Overview · Framework · Method · Release Status · Citation

---

Overview

SatOrtho is a real-time UAV orthophoto generation framework for GNSS-denied environments. Instead of relying on external positioning signals, it uses prior satellite maps as globally referenced cues, aligns them with high-rate visual SLAM poses, and incrementally constructs a georeferenced orthophoto during flight.

Important

This repository is an initial partial release accompanying a manuscript under review. The complete implementation, datasets, and evaluation scripts will be released after acceptance.

Framework

_{SatOrtho combines relative pose estimation, satellite-map-based absolute pose estimation, asynchronous pose fusion, and incremental map reconstruction.}

Method

SatOrtho is organized as four concurrently running modules connected through asynchronous queues.

1. Relative Pose Estimation

Incoming UAV frames are processed by a visual SLAM front end. ORB features, frame-to-frame tracking, keyframe selection, map-point creation, and local bundle adjustment provide high-rate relative camera poses in the visual frame.

2. Absolute Pose Estimation

For each selected keyframe, SatOrtho renders a viewpoint-adapted satellite-map patch and establishes cross-temporal, cross-view correspondences with SuperPoint and LightGlue. The estimated geometric relation anchors the keyframe directly in the geographic coordinate frame without GNSS or GCPs.

3. Asynchronous Relative-Absolute Pose Fusion

A sliding window estimates the similarity transformation between the visual and geographic frames from matched keyframes. High-rate SLAM poses are continuously transformed with the latest estimate, while newly available absolute observations update the alignment without blocking visual tracking.

4. Incremental Orthophoto Reconstruction

New keyframes restore their visible map points, filter elevation outliers, and form a triangulated surface for orthorectification. The resulting image regions are inserted into the global mosaic, where view-angle-aware weights favor near-nadir observations and progressively refine previously mapped areas.

Repository Structure

SatOrtho/
|-- framework.png       # System overview
|-- include/
|   |-- map/            # Mapping and pose-fusion interfaces
|   |-- spdlog/         # Logging support
|   |-- utils/          # Shared data structures and timing utilities
|   `-- zmq/            # Messaging interfaces
|-- src/
|   |-- map/            # Mapping and pose-fusion implementation
|   |-- spdlog/         # Logging implementation
|   `-- zmq/            # Messaging implementation
`-- test/               # Development and integration programs

Release Status

Component	Status
Mapping and asynchronous pose-fusion code	Available
Messaging and supporting interfaces	Available
gis_data implementation	Withheld from the current snapshot
Datasets and configuration files	Planned after acceptance
Build and evaluation scripts	Planned after acceptance

Because required components are intentionally withheld, this snapshot is not yet a standalone buildable release.

Citation

@unpublished{zhou2026satortho,
  title  = {SatOrtho: Satellite-Map-Prior-Assisted Real-Time Orthophoto Generation},
  author = {Zhou, Maoan and Li, Yongfei and Kong, Min and Liu, Jieyu and Luo, Zhaoyun and Yang, Dongfang},
  note   = {Manuscript submitted to IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
  year   = {2026}
}

License

A license will be provided with the complete public release.