ALBERT

Description: The raw images (available on request) have been captured using a Canon 8-15mm fisheye lens and therefore they have a wide field of view, which results in a dark image boundary as the lights did not illuminate the outer sectors well. The images in this dataset have then been undistorted to virtual images that an ideal perspective camera with only 90 degrees horizontal field of view would have seen from the same position. To achieve this, the color of each pixel in the ideal image is obtained by - computing the ray in space associated with this virtual pixel (using rectilinear un-projection) - projecting this ray into the original fisheye image (using equidistant projection), yielding a sub-pixel position - interpolating the colors of the neighboring pixels Technically, the undistortion has been performed using the tool https://svn.geomar.de/dsm-general/trunk/src/BIAS/Tools/biasproject.cpp (at revision 418, and earlier, compatible revisions). Manual image annotation is available here: https://annotate.geomar.de/volumes/262

Description: The cruise SO268 was designed to assess the environmental impacts of deep-sea mining of polymetallic nodules in the Clarion-Clipperton Fracture Zone (CCZ). Therefore, a dredging experiment was conducted on 11 April 2019 between 6:30 to 19:00 UTC. To monitor the dispersion of the generated plume, 15 sensors were distributed around the dredge tracks. Three of them are presented in this study.

Description: Optical imaging is a common technique in ocean research. Diving robots, towed cameras, drop-cameras and TV-guided sampling gear: all produce image data of the underwater environment. Technological advances like 4K cameras, autonomous robots, high-capacity batteries and LED lighting now allow systematic optical monitoring at large spatial scale and shorter time but with increased data volume and velocity. Volume and velocity are further increased by growing fleets and emerging swarms of autonomous vehicles creating big data sets in parallel. This generates a need for automated data processing to harvest maximum information. Systematic data analysis benefits from calibrated, geo-referenced data with clear metadata description, particularly for machine vision and machine learning. Hence, the expensive data acquisition must be documented, data should be curated as soon as possible, backed up and made publicly available. Here, we present a workflow towards sustainable marine image analysis. We describe guidelines for data acquisition, curation and management and apply it to the use case of a multi-terabyte deep-sea data set acquired by an autonomous underwater vehicle.

Description: Images were acquired by the DeepSurvey Camera on board GEOMAR's AUV Abyss. Nodules were delineated by the CoMoNoD algorithm [see related to references]. Result files are computed per AUV dive. Nodule detections below 5cm^2 are neglected as are detections above 707cm^2. Abundance statistics are computed per m^2 and gridded per m^2 as well. For overlapping images, max-pooling has been applied to select the values reported in the result files. Pixel values in the rendered maps correspond to the units reported in the ASCI files (median-nodule-size: cm^2, nodule-number: m^-2, percent-coverage: %, sorting, skewness and pixel-contributions are unit-free).

Description: The latest expedition to the DISCOL experimental area (DEA, www.discol.de) was conducted in 2015 during cruise SO242 with the German RV SONNE. The entire disturbance area was extensively mapped with the ship-mounted Multibeam Echosound (MBES) System. An AUV has been deployed several times in different modes for the acquisition of high resolution MBES, Side Scan Sonar (SSS) and optical data. Even with state-of-the-art positioning technology the remaining navigation uncertainties in the deep sea and the use of a variety of acoustical and optical sensors at different resolutions require detailed post-processing in terms of absolute geographic positioning to improve the overall accuracy of the data. The ship-based bathymetric map of the survey area was used as absolute geographic reference and the AUV-acquired high resolution-data sets available here were geo-referenced and aligned to each other (see the respective assigned PANGEA-links for the different data sets and Gausepohl et al., in review, for details) resulting in the most accurate geo-referenced dataset of the DEA to date. The plough tracks from the disturbance experiment in the DEA in 1989 are visible in all of the acquired high resolution data sets, which enabled digitizing, geo-referencing and age-sequencing of the disturbance tracks (Gausepohl et al., in review). This dataset includes the gridded MBES data from the Kongsberg EM122 mounted in RV SONNE (https://doi.pangaea.de/10.1594/PANGAEA.905579) and the geo-referenced AUV ABYSS-acquired MBES (https://doi.pangaea.de/10.1594/PANGAEA.905580) and SSS (https://doi.pangaea.de/10.1594/PANGAEA.905617) datasets. Furthermore it includes two georeferenced photo mosaics generated from two AUV photo transects (https://doi.pangaea.de/10.1594/PANGAEA.905581). The identified, digitized and georeferenced plough marks are available as GIS-readable shape files here. We further provide the Matrix that was used for the age succession evaluation of the plough marks (Gausepohl et al., in review) in .xlsx and .csv format.

Description: This is the demonstration code for the "Compact Morphology-based Nodule Delineation" (CoMoNoD) algorithm. CoMoNoD is a rapid method to delineate poly-metallic (or manganese) nodules from vertical benthic images. The paper describing the algorithm is currently under review. This algorithm makes extensive use of the OpenCV library for image processing and uses NVIDIA CUDA for computational speedup.