3D scene reconstruction from underwater imagery
Calibration of multi-camera rigs traditionally consists of determination of intrinsic and extrinsic parameters. Effects of refraction that are due to transition of light through interfaces between media with different speed of light are ususlly either ignored or described in terms of lens-related distortions. Physically accurate accounting for refraction requires inclusion of these effects in an image-forming model and determination of refractive parameters, which are very difficult to measure directly.
FAST COLORIMETRIC CALIBRATION OF RGB CAMERAS
Difference in image sensors of cameras does not allow for direct comparison of colors in images acquired by different camera models. Traditional measurement of sensors' quantum efficiency curves is a lengthy and tedious process requiring expensive specialized equipment and laboratory conditions. We are developing a technique for fast measurement of quantum efficiency curves from a single image.
Optimal conditions for calibration of multi-camera rigs
Any quantitative results of image processing require accurate calibration of individual cameras and mutual positions and orientations between cameras in a rig. By means of numerical simulation we investigate optimal configurations of calibration targets and rig structure that lead to highest achievable calibration results.
Probabilistic classification of seafloor using RGB imagery
Color is typically an important cue for classification purposes. However, use of trichromatic cameras underwater often leads to deceiving results due to 1) wavelength-dependent absorption of light by water and dissolved organic matter, and 2) broadband response of camera sensors. We are developing methods to process underwater imagery in a probabilistic manner, utilizing known properties of a camera and spectra of main seafloor sediments.
The overall aim of this tool is to improve data quality issues, to reduce review and acceptance times, and ultimately to reduce ping-to-chart times. Furthermore, once one of the developed algorithms is mature and effective enough, existing commercial software might decide to adopt it with a relatively easy transition based on the existing working implementation.
The Sea Mapper's Acoustic Ray Tracing Monitor and Planning (SmartMap) project aims to provide tools to evaluate the impact of oceanographic temporal and spatial variability on hydrographic surveys. A long-term goal is to calculate a sound speed profiling rate (in hours) to provide a clear suggestion to the surveyor about the timing for execution of sound speed profiles. Such a map would provide guidance to identify problematic spatial and temporal areas to end user surveyors without specialized knowledge.
SOUND SPEED MANAGER
Sound Speed Manager has been designed to ease integration into existing data acquisition workflows. The liberal open source license used by the project (specifically, GNU LGPL) provides for understanding of the chosen processing solutions through ready inspection of the source code, as well as the ability to adapt the application to specific organization needs. This adaptation is eased by the modular design of the application, with the NOAA-specific functionalities organized so that they can be easily deactivated for non-NOAA users.