Researchers at the European Commission project, known as Augmented Reality for Remotely Operated Vehicles (ARROV), investigated a new technique for 3D acoustic image segmentation and modeling to be used in underwater remotely operated vehicles (ROV). The project is a part of the Growth Program, a fifth-framework program of the European community for research, technological development, and demonstration activities. The method uses acoustic devices and exploits the possibility of improving the understanding of a human operator by augmenting the acquired image with a synthetic representation of the observed scenario. The main idea consists of automatic extraction of models appearing on the scene by fitting geometric primitives on acoustic data. The method is based on a simplified version of the so-called recover and select paradigm, in which the seed areas are generated by adopting a robust approach based on random sample and consensus (RANSAC) algorithm. Acoustic data was obtained in 3D form using EchoScope 1600, a high-resolution camera, and the scene was insonified by a high-frequency acoustic pulse. Experiments were carried out to show the effectiveness of the obtained partitions and models, allowing extraction of a great deal of information, useful for improving the perception of an underwater scene.
Efficient scene modeling for underwater ROV driving
Effizientes Verfahren zur Bilddarstellung von Sonarsignalen für die Fernsteuerung von Unterwasserfahrzeugen
Sea Technology ; 45 , 12 ; 47-52
2004
6 Seiten, 2 Quellen
Article (Journal)
English