Aerial Imaging Software: GPU RAW Processing for Drone and Aerial Imaging
Updated:
Aerial and drone imaging produces large volumes of high-resolution RAW frames that have to be processed quickly and consistently. Fastvideo runs the whole imaging pipeline on the GPU — RAW-to-RGB conversion, exposure and shading correction, denoising, tone mapping and image compression — on NVIDIA desktop GPUs and on NVIDIA Jetson for on-board processing. Running it on the GPU turns traditionally offline aerial workflows into real-time ones and keeps color and brightness consistent across a flight, which is exactly what stitching, orthomosaic and photogrammetry need.
Why GPU processing for aerial and drone imaging?
Mapping and survey flights capture thousands of frames per mission, often as RAW to preserve dynamic range. CPU pipelines struggle to process such volumes at capture rate, while in-camera JPG risks unrecoverable over-exposure and a costly reflight. Processing RAW on the GPU is fast enough to preview and cull in real time, applies per-frame exposure correction so every frame in a series shares the same brightness, and produces high-quality RGB that stitches cleanly into an orthomosaic.
The aerial imaging pipeline on GPU
- RAW-to-RGB conversion — high-quality debayer with more key points for photogrammetry, at very high performance
- Individual exposure correction for each RAW frame to avoid over-exposure and keep brightness consistent across a series
- Automatic shading correction (lens/vignetting fall-off) and bad-pixel correction
- Denoising and color calibration
- Adaptive local tone mapping (ALTMapper) — local contrast, shadow recovery, dehazing and soft cloud-shadow removal
- Lossy and lossless RAW image compression and the RawZipper RAW codec to cut storage during and after the flight
- On-board ISP on NVIDIA Jetson for real-time image-sensor control and capture
- Orthomosaic brightness and color alignment across flights — in progress
All of this is available in the Fastvideo Image & Video Processing SDK for integration, and in the Fast CinemaDNG Processor end-user application for RAW workflows.
From capture to orthomosaic: the workflow
For a step-by-step walk-through — from SSD import and culling to per-frame exposure correction and orthomosaic color matching — see RAW processing for aerial imaging. On a modern GPU the RAW-to-RGB stage runs in the range of 1–2 GPix/s, fast enough to preview the whole set as video and check brightness consistency before stitching.
Applications
- Drone and UAV imaging
- Aerial mapping, mobile mapping and street view
- Photogrammetry and orthomosaic production
- Agriculture, inspection and survey
- Geospatial and remote sensing
Roadmap
- Automatic shading correction without preliminary calibration
- GPU-based ISP optimization for aerial pipelines
- Benchmarking of CUDA-based aerial image processing
- Mosaic brightness alignment across all images of a map
Aerial imaging FAQ
What software do I need to process aerial and drone images on the GPU?
Fastvideo provides GPU RAW image processing — RAW-to-RGB, exposure and shading correction, denoising, tone mapping and compression — as the Fastvideo SDK for integration and as the Fast CinemaDNG Processor application, on NVIDIA desktop GPUs and Jetson.
Why process aerial RAW on the GPU instead of using in-camera JPG?
JPG saves storage but an exposure error can make a frame unrecoverable, risking a reflight. RAW keeps the full data, so you can apply per-frame exposure correction in post and keep every frame in a series at the same brightness — and on the GPU it is fast enough to do at capture volumes.
How do you keep color and brightness consistent across a flight and in the orthomosaic?
Per-frame exposure correction equalizes brightness within a series; for the orthomosaic, exposure is corrected per flight and then re-applied to the RAW data in a second pass, so neighbouring images from different flights match. See the workflow article.
Does it run on NVIDIA Jetson for on-board processing?
Yes. The same pipeline runs on NVIDIA Jetson (Orin and earlier) under Linux4Tegra, so RAW acquisition and ISP can run on-board the drone or aircraft.
About the author
Fyodor Serzhenko, PhD, is the founder and CEO of Fastvideo. He earned his PhD at the Moscow Institute of Physics and Technology (MIPT) in 1993. Since 2009 he has led the development of Fastvideo’s GPU-accelerated image codecs and ISP modules, including the Fastvideo SDK. Connect on LinkedIn.
Why you can trust these results
Fastvideo has built GPU-accelerated image processing software since 2009. Our codecs comply with the standards they implement — JPEG (ITU-T T.81 / ISO IEC 10918) and JPEG2000 (ITU-T T.800 / ISO IEC 15444) — and our tools are open source on GitHub. Every performance figure we publish is reproducible: see our benchmark methodology, download the Fastvideo SDK benchmark report (PDF), and measure it on your own GPU.