|
Applied Intelligence 5 (3):207-216, July
1995. © Kluwer Academic Publishers
Multisensor Integration for Underwater Scene ClassificationN.
Nandhakumar Department of Electrical Engineering, University of
Virginia, Charlottesville, VA 22903-2442
nandhu@virginia.edu
S. Malik Department of
Electrical Engineering, University of Virginia, Charlottesville, VA
22903-2442 nandhu@virginia.edu
Abstract We describe a
new approach for the classification of a seafloor that is imaged with high
frequency sonar and optical sensors. Information from these sensors is
combined to evaluate the material properties of the seafloor. Estimation
of material properties is based on the phenomenological relationship
between the acoustical image intensity, surface roughness, and intrinsic
object properties in the underwater scene. The sonar image yields
backscatter estimates, while the optical stereo imagery yields surface
roughness parameters. These two pieces of information are combined by a
composite roughness model of high-frequency bottom backscattering
phenomenon. The model is based on the conservation of acoustic energy
travelling across a fluid-fluid interface. The model provides estimates of
material density ratio and sound velocity ratio for the seafloor. These
parameters serve as physically meaningful features for classification of
the seafloor. Experimental results using real data illustrate the
usefulness of this approach for autonomous and/or remotely operated
undersea activity.
Keywords sensor fusion, feature
extraction, recognition, image analysis, AUV, sonar
ISSN
0924-669X
|