Flir Thermal Model

[Hughsnews]

Hi All,
Having a problem. I am using a Flir Vue Pro to shoot some aerial photographs of archaeological sites. I have seen photoscan be used by other people, but when I try to run it, it just can't seem to produce an alignment. It is saying that the cameras have been aligned and I can get a point cloud of around 60 000 from them using highest settings. It will show points on the screen but not a scene as such, nor camera locations.

The Vue Pro is low res (640 × 512) and I am filming scenes as jpeg files as black/white hot. I have plenty of overlap and am taking hundreds of photos.

Can anyone offer any advice as to settings? I also have a regular photographic model of the same area and I have multiple GCP's. I am just not sure if its the fact that these black/white images are just not different enough, or if I am messing up something in processing as I have seen people creating models from thermal before including in B/W.

Any help would be amazing!

[Packy_GIS]

Hi,
I am playing around with the Optris PI camera (it looks exactly the same like your thermal camera) - I haven't tried Photoscan so far (I will make the survey in two weeks) but here are my suggestions:
- the thermal camera has not lens - its just some kind of sensor
- the resolution is too low for SFM modeling
- it would be hard (or even impossible) to calibrate this "lens"
You should ask yourself a question, if you really need to create orthophoto from thermal imagery (how about the Ground Control Points??) Wouldn't be enough just to use some panorama stitching software and then georeference the resulting image in GIS?
Packy

[sean1192]

Hi Hughsnews,

I've been using the same camera, and have had some success with stitching in Photoscan. I use 'high' alignment, with default key and tie points. Mission settings are usually around 60m altitude, 80% sidelap, and about 90% overlap. Can you confirm that you were able to build the mesh after getting tie points?