Image Processing Reference
we flood the scene with light and look at the reflected energy. A new approach
to acoustic imaging is based on the existence of ambient sound in the underwater
environment. This sound creates a “noise field” that is modified by the presence
of objects in much the same way as scattered sunlight is reflected and absorbed by
objects, making them visible to our eyes. The oceans are full of natural and artificial
sources of noise, including sounds from breaking waves, fish, marine mammals,
crustaceans, and motorboats, to name a few. Much of this noise is above our range
of hearing, but electronic sensors can easily detect it.
In 1985, Michael Buckingham was working on sonar technology for the English
Ministry of Defense when he realized that it might be possible to image objects
underwater with the acoustic equivalent of daylight. Initial studies indicated that an
acoustically reflective object placed at various distances in front of a single-element
detector (a parabolic reflector with a single microphone at the focus) significantly
increased the noise signal level measured by instruments connected to the detector.
This experiment was the acoustic analogue of placing a white sheet of paper against
a dark background in a person's field of view. The Scripps group next designed and
fabricated a larger acoustic “lens” with a cluster of 128 microphones located at the
focus, like the compound eye of an insect. The signal from each microphone maps
to a single “pixel” in the resulting acoustic image.
This group has finally built an instrument containing 900 sensors and extensive
computing power in a flat geometry known as a phased array. 5 The images
it produces are nearly ten times higher in resolution than earlier images. This
instrument, known as ROMANIS, 6 was tested in the warm, shallow seas around
Singapore, where millions of shrimp produce substantial levels of acoustic noise
by the snapping of their claws—a behavior that is not well understood but
serendipitously creates excellent conditions for acoustic noise imaging. The
ROMANIS array is shown in Fig. 5.9.
Figure 5.10(a) shows a visible-light view of a test target imaged by ROMANIS.
The 1-m square aluminum panels are covered with neoprene rubber, which acts
as an effective reflector and diffuser of ambient noise. Figure 5.10(b) shows an
acoustic image of the L-shaped pattern. The pseudocolor indicates the sound
pressure level in decibels, as shown by the color bar to the right of the image. This
image was made with sound in the 58-77-kHz frequency range, which corresponds
to wavelengths in water between 1.9 and 2.5 cm. 7
There are inherent limitations to the resolution of an ANI system, since the
wavelength of the sound waves is a significant fraction of the size of the imaging
system aperture. However, it offers an attractive alternative to sonar imaging, since
5 The phased-array approach eliminates the need for a parabolic reflector and increases the field of
view of the imaging system. The sensors in a phased array detect relative phase (timing) differences
in incoming waves. This information is used to form an image without the need for a focusing lens.
Phased arrays are also used in radar applications, like the passive millimeter-wave camera described
in Chapter 3.
6 ROMANIS stands for Remotely Operated Mobile Ambient Noise Imaging System.
7 1 kHz = 1000 cycles per second.