Home | Contact ST  
Follow ST

Feature Article

Benthic Habitat Mapping Using High-Resolution Image Mosaicking

By Hamed Bagheri
Graduate Student
Engineering and Applied Science

Vincent Lecours
Ph.D. Student
Department of Geography
Memorial University of Newfoundland
Newfoundland and Labrador, Canada

These diagrams show (a) the overlap between sequential image frames and (b) translation and rotation parameters from a submersible's movement affecting image frames and the overlap area.

Benthic ecosystems of most continental shelves, slopes and seamounts are altered by the use of bottom-contact fishing gears and other human activities, such as hydrocarbon drilling and seabed mining. Such activities can impact the structure and function of benthic habitats directly and indirectly.

For example, physical disturbances of the seabed can produce changes in local water flow and sedimentation patterns, which are important for suspension feeders like cold-water corals and sponges. Partial or complete destruction of benthic habitats can reduce the number of species, their abundance, richness and diversity.

In recent decades, efforts have been made to study these remote environments. Mapping seafloor habitats based on species' environmental preferences is often the first step when implementing scientific management, monitoring environmental change and assessing the impacts of disturbance on benthic habitats.

The GeoHab (Marine Geological and Biological Habitat Mapping) international symposium defines benthic habitat mapping as representing physically distinct areas of the seafloor that are associated with a particular assemblage of species. The rapid and recent evolution of technologies has helped provide an almost complete coverage of the seafloor at increasingly higher resolutions.

A new image mosaicking method can facilitate benthic species count recorded from ROV video data. The method creates image mosaics from overlapping images, helping the video data analyst to avoid counting individual organisms twice when an area is surveyed a second time. This improves the accuracy of species distribution maps used to produce benthic habitat maps.

In-Situ Observation of the Seabed
The direct observation of underwater video data allows for an accurate representation of species' distribution and abundance that can improve habitat maps. Knowledge of the geomorphology (e.g., shape of the seafloor) and the surficial geology (e.g., type of substrate) gained from video data collected where specific benthic organisms live helps characterize habitats and test environmental variables that can be good predictors of species distribution.

High-resolution videos also provide relatively accurate views of the seafloor, which can be used to corroborate or complete data collected using other technologies, such as multibeam echosounder systems. Another advantage of the video data is the ability of being analyzed online or offline, and the possibility of extracting image frames if necessary.

A major application of underwater video data in marine sciences is to count and georeference relevant features that are captured in the video, such as biological organisms. Nevertheless, an important issue in using video data for assessing the distribution and abundance of benthic species is the multiple-counting problem, which is counting several times the same feature that appears in several video scenes, a problem that can lead to inflated species counts and inaccurate statistics.

Multiple counting is a problem for machine-reading or automated video analysis, as well as the manual counting process. Assuming that an analyst performed a count properly, there are still conditions that can make the multiple counting possible. For example, a feature can be counted several times if it appears in multiple images or more than once in a video stream.

Overlapping images or videos could happen when the dives are planned for mosaicking or when collecting full coverage of a region of the seafloor. This situation can also occur in dives that had changes to the navigation plan resulting from the topography of the seafloor or scientists requesting exploratory navigation while observing interesting features in real time. These common situations can generate overlaps between sequential images where the submersible is following a route or images taken from adjacent or crossing tracks.

The proposed solution for the problems of multiple counting and overlapping images is to create a photomosaic of the seafloor with imagery data in which each feature only appears once. To continue this article please click here.

Hamed Bagheri has a master's in computer engineering from Memorial University of Newfoundland. He received his bachelor's in electrical engineering from Mazandaran University in 2006. His master's thesis focuses on digital image processing and underwater image mosaicking.

Vincent Lecours is a Ph.D. student working in the Marine Geomatics Research Lab of Memorial University of Newfoundland. He received his bachelor's in applied geomatics from the Universit' de Sherbrooke in 2011. He is studying cold-water coral and sponge habitats in Canada using multiscale bathymetric data analyses.

-back to top-

-back to to Features Index-

Sea Technology is read worldwide in more than 110 countries by management, engineers, scientists and technical personnel working in industry, government and educational research institutions. Readers are involved with oceanographic research, fisheries management, offshore oil and gas exploration and production, undersea defense including antisubmarine warfare, ocean mining and commercial diving.