2009 International Conference on Industrial Mechatronics and Automation
978-1-4244-3818-1/09/$25.00 ©2009 IEEE ICIMA 2009

A Survey of Neuromorphic Vision System
--Biological Nervous Systems Realized on Silicon

Ji-Hong Liu
College of Information Science and Engineering
Northeastern University,
Shenyang110004, Liaoning, China
e-mail: liujihong@ise.neu.edu.cn
Cheng-Yuan Wang and Ying-Ying An
Sino-Dutch Biomedical and Information Engineering
School
Northeastern University,
Shenyang110004, Liaoning, China
e-mail: chengyuan_neu@live.cn


Abstract—Neuromorphic systems emulate the functionality
of neural principles found in biology, especially in the brain
structures of retina and cochlea. The paper focuses on the
introduction of neuromophic vision systems. A brief history
of neuromorphic systems and neuromorphic vision chips is
presented, the advantages and disadvantages of
neuromorphic vision systems are highlighted, and some
research states towards realizing the systems are discussed in
this paper. Finally the development trends of neuromophic
vision systems are summarized. It is shown that the
neuromophic vision systems have promising improvement
prospects.
Keywords-Neuromorphic engineering; vision chip; vision
sensor; address event representation (AER)
I. INTRODUCTION
Compact, efficient electronics based on the brain’s
neural system could yield implantable silicon retinas to
restore vision, as well as robotic eyes and other smart
sensors. -- By Kwabena Boahen
Today’s computers can perform billions of operations
per second, but they are still no match for even a young
child when it comes to skills such as pattern recognition or
visual processing. The human brain is also millions of
times more energy-efficient and far more compact than a
typical personal computer. Neuromorphic microchips,
which take cues from neural structure, have already
demonstrated impressive power reductions. Their
efficiency may make it possible to develop fully
implantable artificial retinas for people afflicted by certain
types of blindness as well as better electronic sensors.
Someday neuromorphic chips could even replicate the self-
growing connections the brain uses to achieve its amazing
functional capabilities [1].
Dr. Carver Mead, an emeritus professor of California
Institute of Technology (Caltech), Pasadena, pioneered the
field of neuromorphic engineering [2, 3]. He reasons that
biological evolutionary trends over millions of years have
produced organisms that engineers can study to develop
better artificial systems. By giving senses and sensory-
based behaviors to machines, these systems can possibly
compete with human senses and bring an intersection
between biology, computer science and electrical
engineering [4]. Neuromorphic vision sensors and
preprocessors are increasingly being used to implement the
first steps of visual processing in artificial systems. A
typical application domain is autonomous mobile systems,
for which size, power consumption and speed are
important parameters. While until recently research in
neuromorphic vision concentrated on designing and
optimizing individual circuits, part of the effort is now
shifting towards multichip neuromorphic systems and
hybrid systems interfacing neuromorphic circuits with
other types of processors [5].
II. ANALYSIS OF ADVANTAGES/DISADVANTAGES IN
VIEW OF THE RECENT PROGRESS
A. The Advantages of Neuromorphic Vision Systems
Neuromorphic technology, especially neuromorphic
vision sensors belong to a new type of electric circuits that
prove to be very useful in the domain of mobile behavior
based robotics. Neuromorphic technology emulates the
functionality of neural principles found in biology,
especially in the brain structures of retina and cochlea [6].
1) Control the chip in real time
A human brain must process sensory information in
real time in order to analyze its surroundings and prescribe
appropriate actions. However, most simulations of neural
functions now have been executed in software programs
that run much more slowly than the required speed of real
time. Neuromorphic hardware aims to emulate the
functionality of the brain, using silicon analogs of
biological neural elements (Mead, 1989). Typically, unlike
most of the software programs, hardware models can
operate in real time (or even faster than their biological
counterparts), providing the opportunity to create artificial
nervous systems that can interact with their environment.
Reconfigurable neuromorphic systems represent a
compromise between the fast, dedicated silicon hardware
and the slow but versatile software. Therefore, the systems
are useful for learning the real-time operation of high-level
(e.g. cortical), large-scale neural networks and prototyping
neuromorphic systems superior to fabricating application-
specific chips [7].
2) Implantable microchips
Nowadays high-performance implantable electronic
devices are commonly used in the biomedical domain, and
several biomedical electronics are designed for neuro-
prosthesis applications. These implantable electronics
devices include cardiac pacemakers, cochlear prostheses,
retinal prosthesis systems [8], and functional
neuromuscular stimulation systems. For the nervous
154