Calibration of refractive housings for underwater vision

Abstract Many underwater applications rely on vision sensors and therefore require accurate camera calibration, i.e., precise knowledge of the incoming light ray associated with each image pixel. While in the ideal pinhole camera model all viewing rays intersect at a single point in 3D space, underwater imaging systems are affected by potentially multiple refractions at the interfaces between water, glass, and air. These changes in ray direction depend on the camera’s position and orientation within the waterproof housing, as well as on the geometry and optical properties of the interface window (port). In recent years, explicit models for underwater imaging through commonly used ports, such as flat and dome ports, have been proposed. In our previous work, we introduced the open-source tool Calibmar for refractive underwater camera calibration. In this paper, we provide further implementation details of this toolbox and analyze the magnitude of errors that arise when refractive effects are neglected.