A novel monocular calibration method for underwater vision measurement

Vision measurement systems have a reliable performance on ground, but it remains a challenge to apply commonly-used vision measurement systems (i.e. multi-camera systems and laser systems) in underwater environments. One of the most challenging issues is the transformation from an unscaled measurement to a scaled result, which is achieved by a calibration method and determinate the strategy used for underwater vision measurement. This paper proposes a novel monocular underwater calibration method underlying a simple underwater vision measurement system. Underwater unscaled measurement results are calculated by the dark channel prior model. These results are then processed by our calibration method, transforming the unscaled measurements to accurately scaled results. These measurement results finally are used to estimate the scaled 3D structure of underwater objects. Experimental results under natural open water show that our proposed method is reliable and efficient.