Advances in fusion of high resolution underwater optical and acoustic data

We report efforts to merge data from the complementary modalities of optical and acoustic sensing for obtaining more accurate representations of the seafloor. We show that the principal obstacles to merging the acoustic and optical imaging modalities are the distortions inherent to each modality. The construction of geometrically accurate photomosaics is dominated by incremental errors arising as individual images are scaled and warped to form the photomosaic. For microbathymetric mapping, principal errors arise from sensor position and orientation calibration parameters that affect our ability to construct maps from sonar data that are commensurate with sensor and navigation resolution. We show that photomosaics can be combined with microbathymetry through simple, locally defined, finite element warps. An error analysis quantitatively evaluating photomosaic geometric distortions is reported. This technique is shown to be successful in providing opto-acoustic high resolution three dimensional perspectives of large areas of unstructured seafloor features. These issues are evaluated with a combined optical and acoustic survey data set obtained from a 675 kHz pencil-beam sonar and electronic still camera survey of a 4th century B.C. Roman shipwreck. This survey was conducted by the authors with the Jason ROV in the Mediterranean Sea in 800 meters of water depth.