PUMA
Istituto di Scienza e Tecnologie dell'Informazione     
Reggiannini M. A cooperative approach for pattern recognition in underwater scene understanding by multi-sensor data integration.
 
 
Abstract
(English)
Marine environments cover more than two thirds of the whole earth surface and represent an outstanding scenario for researchers and scientists. Concerned people are involved in specifi c topics such as marine biology, oil and gas extraction, environment pollution monitoring and cultural heritage safeguard and preservation. This peculiar environment represents a hostile framework for what concerns human or scientifi c operations, in particular archaeological surveys or intervention missions. Oceans' waters impose strict constraints to any kind of underwater activity, included survey, mapping, rescueing and manipulation of sunken objects. Nevertheless the world seas have become quite a concern to cultural institutions since the number of wrecks lying all over the globe seabeds has been estimated by UNESCO to be around 3 millions. This huge patrimony is currently threatened by criminal activities which have tools available to discover underwater sites and illegally remove their content. Joint e fforts between cultural institutions and scientifi c communities have been fostered by the European Community to promote survey missions of the marine seabeds and safeguard actions aiming at the preservation of the archaeological sunken heritage. To face the complicated issues concerning any kind of human activity in the peculiar marine framework, technical operators and useful support in the devices typically used in the oceanography fi eld. Surveys can be performed by unmanned robots and this enables efficient data capture campaigns to be carried out and allows the thorough and detailed observation of extremely remote locations, including those too deep to be easily accessed by human operators. Many typologies of robots have been devised for exploitation by oceanographers. They are usually classifi ed as Remotely Operated Vehicles (ROV), semi-autonomous platforms requiring human operators to be maneuvered, and Autonomous Underwater Vehicles (AUV) that can be programmed to perform survey missions in a completely autonomous mode. The experimental missions carried out within this PhD activity have been performed by exploiting AUVs designed and developed in the framework of the 7th European Framework Programme project "ARchaeological RObot Systems for the World's Seas". For the project purposes, the robots have been equipped with a set of payload sensors, properly selected bearing in mind the specifi c mission requirements. Optical and, most notably, acoustic sensors are the natural choice to survey the sea environment. Acoustic sensors are particularly appealing because of the remarkable efficiency of acoustic propagation in the water medium. Actually acoustic waves propagate over long distances in the water and may warrant signifi cant coverages (hundreds of meters or more) without su ffering from strong energy loss. Among the Sound Navigation and Ranging (SONAR) devices, we mention the Side Scan Sonar (SSS) and the Multibeam Echosounder (MBES). The former generates maps of the seafloor, providing the operator with a large-scale but coarse-resolution visual representation of the environment. The latter returns bathymetry information of the seafloor as well. Depending on the sensor typology the spatial resolution properties may vary from meter to centimeter values. This thesis addresses the development and implementation of methods and procedures aimed at providing a robot platform with an autonomous capability to understand the marine environment without human supervision. Each sensor onboard generates a data stream and through proper processing returns a piece of information concerning the environment as seen from the specifi c viewpoint of the considered sensor. The fusion of the multiple pieces of information not only facilitates the acquisition of an overall picture of the environment but also allows to achieve increased reliability about the recognition of the objects in the scene and about its interpretation.
Subject Underwater object detection
Optical and acoustic data processing
Mosaicking
3D reconstruction
Shape recognition
Texture analysis
I.4.3 IMAGE PROCESSING AND COMPUTER VISION. Enhancement
I.4.6 IMAGE PROCESSING AND COMPUTER VISION. Segmentation
I.4.7 IMAGE PROCESSING AND COMPUTER VISION. Feature Measurement
I.4.8 IMAGE PROCESSING AND COMPUTER VISION. Scene Analysis
I.5.4 PATTERN RECOGNITION. Applications
68T45 Machine vision and scene understanding


Icona documento 1) Download Document PDF


Icona documento Open access Icona documento Restricted Icona documento Private

 


Per ulteriori informazioni, contattare: Librarian http://puma.isti.cnr.it

Valid HTML 4.0 Transitional