As I near the end of my doctoral studies, I reflect on how different my thesis is from what I actually started four years ago but at the same time how much things come full circle. My involvement began when I started using UBC-Gavia, an Autonomous Underwater Vehicle, to map the bottom of the lake. Unfortunately, as a result of the slope steepness in this lake, we found it very hard to accomplish and so the focus of my thesis is on water temperature and physical transport. That said, I’ve maintained a soft spot for image mosaicing.
Just recently, we have been working with people from the Center for Coastal and Ocean Mapping (CCOM) and the University of Delaware to mosaic not only the images we have been collecting but also those Deepworker images. The first, and easier dataset to work with, was the flat sections in the middle of the lake which has been of interest due to the microbialite mats that have been observed there. These are easier to process as don’t have roll and pitch errors that are introduced. Below is just a very small sample of what the final product that can be generated.
In addition to running AUVs, I am also lucky enough to participate in PLRP by being a Deepworker pilot and I was able to have my first flight yesterday. After finishing my mission yesterday and completing all my objectives, I was told that I had a bit of extra time left over so I leaped at the opportunity at testing my new found mosaicing skills. As I was coming back to the barge, I passed by what people around here call ‘microbiliate roads’; long straight lines of microbialite that are aligned along the slope. Lining up the camera, I tried to film a long straight line up the slope. Although the mosaic still has some error resulting from vehicle pitch – you can see this in the image by the fact that it begins to ‘pinch’ out – but I was still pretty happy with the first attempt.
So now the next step is to refine the processing so that we can start using these images for our mission planning for both AUV and Deepworker flights. Part of doing this is to clean the images to remove the roll and pitch effects and then we can drape these images over the bathymetry data that we are collecting. This will allow us to start creating a georeferenced map of the photos.