Automatic image analysis of sediment types : mapping from georeferenced video footage on the Labrador Shelf / K. Jerosch ... [et al.]. : M183-2/6752E-PDF

This Open File describes mapping sea floor features efficiently from video footage providing a guideline for every required step. This is achieved by 1. developing a method to extract single frames from the video footage and make them suitable for Geographical Information Systems (GIS) (georeferencing the frames automatically), and 2. applying the Geospatial Image Database and Analysis System GIDAS (Lüdtke et al., accepted) on HD video frames from the Labrador Shelf, GIDAS provides an approach for a fully automatic detection and the quantification of sediment texture on sea floor images/frames on the Labrador Shelf. This is performed based on the extraction of textural features from the images and a statistical classification of these features using machine learning techniques. The approach includes a quality control based on a cross validation of the predictions. 3. Furthermore, the mapping of the results and discussing GIDAS of being a suitable technique for sedimentological seabed mapping forms the concluding scope of the Open File report.

Lien permanent pour cette publication :
publications.gc.ca/pub?id=9.821643&sl=1

• Format MARC XML
• Format MARC HTML

Demander des formats alternatifs