In this project we propose the development new algorithms and techniques for the automatic construction of ocean floor photo-mosaics. Photo-mosaics allow the geological analysis of ocean floor morphology and hydrothermalism of selected areas. Likewise, underwater photo-mosaics provide high quality images of specific areas for the study of biological communities.

Garbi AUV The works which have been carried out up to now in the field of automatic construction of photo-mosaics have been limited to processing planar environments, while most of the regions of interest for the scientific community are located in areas with 3-dimentional relief. For this reason, research and development of new algorithms is a must to enable simultaneous photo-mosaicking and estimation of underwater 3D relief. On the other hand, none of systems described in the literature is able to automatically build large-scale photo-mosaics, due to the accumulated drift as the mosaic increases its size. In this project we propose to develop new "sensor fusion" strategies to take advantage of all the information available from the on-board sensors to automatically build bigger and more reliable photo-mosaics.

Finally, the proposed system should be modular and flexible, to adapt to a wide diversity of underwater vehicles equipped with various types of sensors. Such a system will allow the analysis of the temporal evolution of both existing ecosystems and morphology of the underwater terrain.