Abstract
Underwater visibility most certainly impacts a scientific diver’s ability to collect high quality images and photogrammetry models in a subjective sense. My project aimed to better understand if varying visibility affects photogrammetry model quality in a more objective manner. The overall purpose of my study was to help scientific divers make better, more efficient use of their time both when collecting photogrammetric data underwater and when processing data in a laboratory environment. Though non-DSLR camera methods for underwater 3D model creation do exist (LiDAR, sonar, etc.), possessing a better understanding beforehand of which environmental visibility conditions would be greatly beneficial. Better quality data leads to higher quality orthomosaic and point cloud models. This advance knowledge can save time during the time-consuming, data-intensive processing phase by, for example, only processing data that meet a certain visibility threshold and archiving data that don’t meet that threshold. Alternatively, dive expeditions can be postponed entirely if visibility does not meet a necessary threshold. Quality was measured in terms of spatial accuracy or the ability for software (Agisoft Metashape and MATLAB pclt) to accurately measure object dimensions. I hypothesize that as visibility decreases, so too will image quality and models, thereby resulting in decreased spatial accuracy overall. At the photogrammetry lab within the University of Miami Physics Department and Rosenstiel School of Marine, Atmospheric & Earth Science (RSMAES), I joined the team in conducting data collection dives primarily around Bear Cut Bridge and occasionally at nearby reef sites such as Rainbow Reef. The control group items used for imaging purposes were unexploded ordnances (also referred to as UXO or munitions) supplied to RSMAES by the United States Navy’s Strategic Environmental Research and Development Program (SERDP). After examining measurement accuracy alongside underwater visibility, a direct relationship was discovered with increasing or decreasing visibility leading to higher or lower accuracy respectively thereby confirming my hypothesis.