Manovich, Lev - The Engineering of Vision from Constructivism to Computers.pdf

(807 KB) Pobierz
Microsoft Word - ev_fin.doc
Lev Manovich
The Engineering of Vision from Constructivism to Computers
INTRODUCTION
1
CHAPTER 1. VISUAL ATOMISM
8
1. I NTRODUCTION 8
2. E XPERIMENTAL P SYCHOLOGY AND THE S CIENCE OF A RT 11
3. V ISUAL A TOMISM : A C ODE FOR M ASS C OMMUNICATION 16
4. V ISUAL A TOMISM AND THE M IND -B ODY P ROBLEM
22
5. V ISUAL E SPERANTO 27
6. C ONCLUSION
31
CHAPTER 2. I SEE, THEREFORE I THINK
35
1. I NTRODUCTION 35
2. I S EE , T HEREFORE I T HINK 39
3. F IRST S IGNS OF R EVOLT : V ENN AND G ALTON 50
4. "T O TEACH THE WORKER TO THINK DIALECTICALLY ."
54
5. F REUD ' S T HEORY OF V ISUAL R EASONING
61
6. T HE R ISE OF THE D IAGRAM 68
7. V ISUAL T ECHNOLOGIES AND THE M IND 77
8. A NALOG E NGINE
85
CHAPTER 3. MAPPING SPACE
90
1. V ISUAL N OMINALISM 90
2. "T HE MOST IMPORTANT EVENT OF THE R ENAISSANCE ." 99
3. R ADAR : S EEING W ITHOUT E YES 115
4. 3-D C OMPUTER G RAPHICS : I NTERACTIVE P ERSPECTIVALISM
125
5. C OMPUTER V ISION : A UTOMATION OF S IGHT
132
6. C ONCLUSION
146
CHAPTER 4: THE ENGINEERING OF VISION FROM INKHUK TO MIT
150
1. N OT A RTISTS BUT E NGINEERS 150
2. I NFORMATION T HEORY : AN E NGINEER A NALYZES C OMMUNICATION 156
3. T HE I NFLUENCE OF I NFORMATION T HEORY OR THE I DEOLOGY OF THE C ODE 162
4. F ROM "H UMAN M OTOR " TO "H UMAN I NFORMATION P ROCESSING " 173
5. C OMMUNICATION E NGINEER A NALYZES H UMAN V ISION 177
6. H UMAN E NGINEERING 186
7. C ONCLUSION : THE L ABOR OF P ERCEPTION
193
CONCLUSION 196
BIBLIOGRAPHY
199
FIGURES
209
Introduction
The dissertation presents a history of modern ideas about vision. I believe that vision is
not a timeless concept; rather, each period understands vision differently depending on how it is
used. In the twentieth century, vision acquired new roles as the medium of mass communication
and the instrument of labor, and, as any other productive tool, it was subjected to engineering,
rationalization and automation. Such new disciplines as applied experimental psychology and
cognitive science, communication engineering and film, robotics, and advertising design
continue to search for ways to utilize vision productively. In the process, they generate new
knowledge about vision, at the same time reducing it to a few disjoined and limited models. The
dissertation chapters follow the development of four such models: vision as a code, vision as a
means of logical reasoning, vision as a way to capture spatial information, and vision as
information processing.
Let us consider a few definitions of vision that are representative of entire research paradigms
and that were unthinkable before the middle of this century.
David Marr's Vision, published in 1980, summarized a decade of investigations on
human perception carried out at the MIT Artificial Intelligence Laboratory. This book has been
the most influential account of the computational approach to vision, shared by computer
scientists and psychologists. It opens with this statement:
What does it mean, to see? The plain man's answer (and Aristotle's, too) would be, to know what
is where by looking. In other words, vision is the process of discovering from images what is
present in the world, and where it is. 1
1 David Marr, Vision (New York: W.H. Freeman and Company, 1982), 3.
 
There is nothing "plain" about this definition of vision; it is functional and pragmatic. Visual
perception is reduced to a number of computational processes for the recovery of limited
information about the world: the identity of objects and their positions. In fact, this is the only
kind of information which may be required for a robot or an automatic missile to perform its
task. Marr projects these goals of machine vision onto human vision. Here, vision is reduced to
the common denominator shared by humans and low level organisms: to detect an obstacle, a
predator, a prey.
In a special 1984 issue of Cognition (the leading periodical of cognitive science) devoted
to visual representations, Steven Pinker outlined the understanding of vision commonly held by
cognitive psychologists:
Certain abstract problems could be best solved by translating their entities into imagined objects,
transforming them using available image transformations, detecting the resulting spatial relations
and properties, and translating those relations and properties back to the problem domain. 2
In this definition, vision is valued merely for its topological properties -- the ability to represent
such relations as inclusion, proximity, and relative positions. According to many cognitive
scientists, these properties make it a more efficient medium for problem-solving and abstract
thinking than language. And since the mind itself is imagined as an exemplary model of
computational efficiency that evolved through evolution, scientists postulate the centrality of
visual (read: topological) representations for various mental processes. To think productively and
without waste, we might dispense with language and instead think through images, submerging
ourselves into the silent movie theater of our minds.
Here is another model of vision that narrows its definition to a particular productive
property. In a 1951 overview of applied experimental psychology, the field which is today
known as human factors, Paul Fitts writes:
2 Steven Pinker, "Visual Cognition: An Introduction," Cognition 18 (1984): 66.
 
Each sense modality has certain inherent advantages and disadvantages for the detection and
analysis of different kinds of information. Audition is more nearly a continuous sense than vision;
vision is basically selective and intermittent. As a consequence, audition is well adapted for the
detection of warning stimuli that may arise at any moment from one of a variety of sources,
whereas vision is well suited to the selection of and concentration on particular stimuli to the
exclusion of others. 3
What is the purpose of this research into the relative properties of the senses? In contrast to a
manual worker, the job of an operator in a modern human-machine system, be it an airplane
cockpit, radar display or an automated production line, is primarily perceptual in nature. His or
her work is to monitor the displays and to detect those signals that require intervention. Starting
in World War II, experimental psychologists have collaborated with engineers in the design of
displays and perceptual strategies for their operators. One of the important questions in this
research has been the relative advantages of visual, auditory and tactile displays for the detection
and analysis of different kinds of signals. Consequently, Fitts describes vision as a sense which is
most reliable and efficient for the constant surveillance of a single source -- it is "basically
selective and intermittent." According to this model, in order to use vision most productively,
one's eyes must be literally "glued to the screen."
In each of these models, vision is understood in terms of its efficiency in performing
specific tasks: recovery of three-dimensional information, logical reasoning, detection of signals.
In fact, there is no single meaning of "vision" which is shared among them -- "vision" as such
does not exist. The only thing that unites these modern approaches to vision is the quest for the
efficient, reliable, and effective instruments of labor. In this respect, vision is just a set of
separate tools which can be employed by either a human or a machine to get work done.
It is this emphasis on the productive uses of vision that distinguishes my project from a
number of recent histories of vision. In 1988 Hal Foster edited the anthology Vision and
3 Paul Fitts, "Engineering Psychology and Equipment Design," in Handbook of Experimental
Psychology , ed. S.S. Stevens (New York and London: John Wiley & Sons, Inc., 1951), 1314.
 

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika:

Zgłoś jeśli naruszono regulamin