Full Motion Video: Issues with altitude and frames with undersea data.

[GraceFrontin-Rollet]

Hi everyone, 

I have come across and interesting issue and was hoping that someone else may have managed to trouble shoot this. I am multiplexing undersea video imagery and metadata to hopefully create some mosaic images of the seafloor. 

Every time I run the multiplexed video, there are random areas where the true sensor altitude goes from -880 (correct) to 18000 m, for example. The video frame is also doing some very strange things and getting very large and then shrinking to the correct size.

The strange thing is, I have visited the timestamps in the metadata file, and the altitude is correct. There have been no sudden changes in other metadata types used. I have also tried multiple DEMs (ESRI Imaging service bathy, some in house high res bathy, and no DEM at all). I have even checked the file formatting and set everything to numerical in case it was just a small formatting issue.

The sensor altitude is only ~1-5 metres above the seafloor, so if the reported water depth is -800m, then the sensor altitude is calculated to -795m. From all the information I could find this seems the correct way to do this.

Unfortunately, due to the nature of this work I cannot share screenshots/data. 

I am using ArcGIS Pro 2.6.1.

Any ideas are greatly appreciated.

Thanks.

[GraceFrontin-Rollet]

I have just solved this problem myself, and thought I'd post what I've figured out in case anybody else has this problem.

1. The altitude limits in the multiplexer tool are -900 for bathymetry. My altitudes went below this so it multiplies itself, hence -900 x -900 = 18,000.

2. A work around is to add 300m to every altitude value AND use the raster calculator to add the same to your DEM layer.

3. Ensure you set your transformed raster DEM to the map elevation source, and the elevation source in the multiplexer tool.

After this your relative altitude between your video frames and DEM will be correct, and the size of your footprint will also be correct. But do remember that anything extracted from this will then need to be backwards corrected for the altitude added in the first steps; I'm still figuring out the backwards calculation steps.

[graharg_TCarta]

Thanks for sharing.  I hope you can share some of video imagery someday.