Conditions of Underwater Image Acquisition for Optimal 3D Reconstruction

Yuri Rzhanov and Igor Kozlov


Friday, Feb. 17, 2017, 3:00pm
Chase 130

3D reconstruction of underwater objects can be achieved with submillimeter accuracy only by means of optical imaging. We discuss design of a multi-view system with conventional cameras. Cameras' and underwater housings' specifications and system configuration define quality of reconstruction. We have developed a simulation framework that allows the modeling of the effects of various parameters on the reconstruction result. For verification of simulation results, a multi-camera rig has been built. In-air experiments support the conclusions derived from the simulations. Experiments with a camera in an underwater housing have shown an importance of refraction of light on both interfaces: air-acrylic and acrylic-water. Attempts to disregard this effect or to describe it within the model of radial lens distortion, as it is often done, may lead to significant errors in reconstruction. We also discuss directions of future research and possible applications.


Yuri Rzhanov
Dr. Rzhanov's research interests are in the signal and image processing and analysis. Specifically, underwater optical sensing, probabilistic reconstruction of object’s color in attenuating media, 3D reconstruction of scenes, automatic monitoring and classification of aquatic habitats, estimation of species’ abundance, classification of seafloor facies from acoustic backscatter. After defending PhD thesis on nonlinear self-organization phenomena in solid state materials, he worked on numerical modeling of nonlinear effects in lasers, development of interactive educational software, and 2D and 3D signal processing. Currently Dr. Rzhanov is a Research Professor at the Center of Coastal and Ocean Mapping, UNH, and is working on classification of marine habitats by means of optical sensing.

Igor Kozlov
Igor Kozlov received his B.S in Information Systems and Technologies from Saint Petersburg State Polytechnic University. He is currently pursuing a master’s degree in computer science UNH. His research interests include image processing and underwater image reconstruction.