Wide area 3D seafloor reconstruction and its application to sea fauna density mapping

3D visual mapping of the seafloor has found applications ranging from environment monitoring and survey of marine minerals to underwater archeology and inspection of modern man-made structures. However, the attenuation of light is significantly more pronounced in water than in air or in space, and so in order to obtain underwater images in colour, it is typically necessary to be within 2 to 3m of the seafloor. In addition to the high risk of collision when operating underwater vehicles at such low altitudes, the limited area of the seafloor covered in each image means large area surveys require a huge investment of time. In this research, we aim to increase the efficiency of mapping large areas of the seafloor by developing an underwater imaging system that can take colour images at ranges of up to 12 m, so that each image can cover a larger area, together with the necessary algorithms to automatically process the data it obtains. The system was deployed to map artificial hydrothermal vents in Iheya North Knoll using the ROV Hyper-Dolphin in October 2012. In this paper, we describe the methods used to process the data it obtains and present wide area 3D reconstructions of habitats surrounding artificial hydrothermal vents. Habitats of Shinkaia crosnieri squat lobsters, which are abundant in the hydrothermally active areas, are identified in the maps and their population density calculated.