In the interest in maximizing my ability to cover lots of acreage per flight, and I don’t need a beautiful looking ortho or DSM, is there a process to simply georeference (that’s the term that comes to mind, but also could be stitching, mosaicing…not sure) images with only like 5% front and side overlap, or even no overlap? I don’t care if trees are falling over, I just want a georeferenced raster dataset where I can look at all the images in a GIS - preferably as a mosaiced tif or something like that. I am trying to conduct broad area imagery searches, that’s why I don’t need ortho quality, just looking for particular objects. The quantity of data precludes manually georeferencing. I also can’t just drop the raw jpg into a GIS and have it lay where it needs to.
Give it a try with Microsoft Image Composite Editor or another naive stitching program (Hugin) and see if they can make anything of it. If not, you’re probably out of luck.
It’s not an issue of Metadata, more of doing the reconstruction. Matching features isn’t easy, which is why sufficient overlap is necessary. The more pixels in common, the more likely a feature can be matched.
A robust naive stitching program doesn’t need/use the Metadata at all, just matched fearures, and may be more tolerant of the low overlap.
I guess what i meant about metadata is the sensor has a GPS position of varying accuracy, it knows technical specs about it’s sensor and where it’s pointing, I would have thought it could then determine coordinates for the four corners of an image.