Signal Processing - September 2017 - 3

FROM THE EDITOR
Min Wu | Editor-in-Chief | minwu@umd.edu

Cameras, Music, and Synergy in Signal Processing

hen you receive this issue of
IEEE Signal Processing Magazine
(SPM), the 2017 edition of the
IEEE International Conference on Image
Processing will soon be taking place in
Beijing, China. This is the first time this
flagship conference of the IEEE Signal
Processing Society (SPS) will be in
mainland China, marking a milestone
of more than five years of strong effort
by the SPS to strengthen membership
development and engagement in China
(which includes three annual summits
and conferences of ChinaSIP in 2013-
2015, followed by a successful ICASSP
held in Shanghai last year).
The digital camera is perhaps one of
the most tangible impacts of image processing that can be felt by an ordinary
citizen. By just counting the number of
pixels per U.S. dollar of cost, one could
already appreciate the paradigm-shifting advances of the underlying technologies. In 2000, the first consumer-level
digital point-and-shoot camera that I put
my hands on was a 1-million-pixel unit
that my research group received from an
industry collaborator. The manufacturer's suggested price was US$1,000,
excluding memory cards, which cost at
least another $US100 for just 32 megabytes. Today, a decent digital camera
over 10 million pixels is priced below
US$10 0. Another indicator was an
informal poll on digital versus film cameras. Even in the mid-2000s when price
Digital Object Identifier 10.1109/MSP.2017.2730938
Date of publication: 6 September 2017

had significantly dropped and image
quality and features had improved, if
you asked an amateur photographer,
more often than not he/she would favor
a film camera, citing its dynamic range,
naturalness, richness in color, among
other qualities. It was not uncommon
to hear their prediction that digital may
never catch up with film technologies.
Today, however, even serious professionals by and large have adopted the
use of digital cameras. Handling higher
dynamic range, providing color calibration, and having nearly imperceptible
effect from superb compression are just
a few of the advances contributed by
the signal processing community. If you
haven't already done so, check out the
September 2016 issue of SPM featuring
a special section on computational photography and display.
A similar "never catch up" statement
regarding digital technologies can be
found in another form of artistry-
musical instruments, such as the piano.
A digital piano today has progressed
from the electronic keyboards you may
have seen in a toy store. In many ways,
it is trying to mimic a traditional acoustic piano without having physical strings
being struck by hammers and resonated
by a soundboard. From the sensors used
to record the manner of a key being
pressed by a player (known as touch), to
music synthesis, to storing the music in
compact files for archival, signal processing plays a critical role in this mimicking game. Traditional piano teachers
tend to almost habitually favor acoustic
IEEE SIGNAL PROCESSING MAGAZINE

September 2017

pianos, yet increasingly many of them
praised the touch and sound of highend digitals in blind tests. Perhaps
the paradigm shift from the acoustic to
digital piano is already starting, despite
at a slower pace compared to the takeover of film cameras by digital cameras. This may be partly due to the
durable nature of pianos and the price
point of about an order of magnitude
higher than cameras.
Among the various modules in a
digital piano, touch is perhaps the hardest to achieve perfection. One of the
leading design frameworks is by industry giant Yamaha. Yamaha approached
the touch problem in its high-end products by implementing a hybrid design
to take advantage of the best in both
the physical and digital worlds: in particular, it has retained the mechanical
design of the actual keyboard part (yes,
you read it right: the keyboard part is
the same as Yamaha's acoustic grand
piano). Instead of striking a string, the
key action will be sensed by multiple
optical sensors and sent to a central
processing unit to analyze the fashion
of each key action. In highly simplified terms, the analyzed key action
will guide a comparison of the typical
ways a key would sound on Yamaha's
best grand piano by its corresponding
touch and then render into the actual
sound by sophisticated sound synthesis
and projected out by a speaker system.
This is just one of the many examples
showcasing the advances in music signal processing.
3

Table of Contents for the Digital Edition of Signal Processing - September 2017