Ground Control Point Editor

Project Goals

• Load TIFF imagery, either via COG URL, or uploading a file.
• Load GCP coordinates from a file.
• Pinpoint on the tiff imagery exactly where the GCPs are located.
• Output a GCP .txt file for use in ODM and other tools (this file links real world coordinates to x/y pixel coordinates on an image).

Implementation

• The most suitable candidate for this is a Web Component.
• A Web Component can be standalone, easily embedded anywhere on the web.
• We have an image georeferenced by a drone in EPSG:4326.
• Ideally we need to reproject EPSG:4326 to a cartesian UTM projection, allowing us to accurately extract pixel coordinates from the image and pair them with lat/lon coords in real life.
• By far the best candidate for this is OpenLayers, the mapping library with the best projection support.
• We may also need to use geotiff.js for low level calculations / math to achieve what we need.
• An example using both can be found at https://github.com/geotiffjs/cog-explorer

Note that MapLibre has a new COG protocol extension, but this only supports EPSG:3857 images and we do not wish to introduce that inaccuracy when dealing with precise GCP.

Related forum post that will influence our decisions here: https://community.opendronemap.org/t/passing-different-output-projections-to-odm-eg-a-proj-flag/22460

Final Step

• In the end we have the GCP .txt file and all photos in EPSG:4326 projection.
• These files are sent to NodeODM for processing via ODM 'split merge' mode.
• ODM decides on best way to divide up the photos for efficient processing.
• The GCPs are used to georeference the images.
• The images are combined into one large orthomosaic.