Test Structure From Motion UnderwaterLatest update July 22, 2015 Started on June 18, 2015
On Tuesday July 14th, we took an ROV to a sunken sailboat in Lake Tahoe. I made multiple transects over the boat, and a GoPro mounted downwards took photos twice a second. The photos produced a rough 3D model, but a second trip will probably be required to fill in the gaps. Here is one of the approximately 10,000 photos taken by the GoPro during the dive.
Today we are going to attempt to map a portion of Lake Merritt. We will take video of the lakebed with a downward facing GoPro. To turn the video into a 3D model, individual frames will be post-processed in VisualSFM and Meshlab. I have already experimented with creating a model from video. The image below shows a model of the ground created from video recorded with my phone. I used GOM Player's Burst Capture to get a sequence of frames, and I then followed an Open Source Photogrammetry tutorial to turn the images into a 3D model.
Contribute to this expedition
Thank You for Your Contribution!