|
We
are in the midst of a revolution sparked by rapid progress
in digital image processing technology. Hareish Gur
explains the nuances of Image Processing and looks at the
range of applications in which the technology is being deployed
 Image
Processing is considered to be one of the most rapidly evolving
areas of information technology today, with growing applications
in all areas of business. This technology holds the possibility
of developing the ultimate machine in the future that would
be able to perform the visual functions of human beings. It
also forms a core area of research within the computer science
and engineering disciplines at most of the top universities
and institutes in the US and other developed countries. As
such, it forms the basis for all kinds of future visual automation.
Image Processing deals with images which are two-dimensional
entities (such as scanned office documents, x-ray films, satellite
pictures, etc) captured electronically through a scanner or
camera system that digitises the spatially continuous coordinates
to a sequence of 0’s and 1’s. A digital image is
a mapping from the real three-dimensional world to a set of
two-dimensional discrete points. Each of these spatially distinct
points, holds a number that denotes grey level or colour for
it, and can be conveniently fed to a digital computer for
processing. Here, processing essentially means algorithmic
enhancement, manipulation, or analysis (also understanding
or recognition) of the digital image data. Every image processing
technique or algorithm takes an input, an image or a sequence
of images and produces an output, which may be a modified
image and/or a description of the input image contents.
For example, if we give as input, a person’s photograph,
an image processing system could return his or her name and
whether or not that person is wearing glasses or a necktie.
Since a digital computer is used in the process rather than
an analogue one, this branch of studies is also popularly
known as Digital Image Processing.
Importance of image data
According to one estimate, more than 75 percent of all the
information received by man is visual. Some researchers arguably
consider this figure to be as high as 99 percent! Even if
we consider the conservative estimate, the remaining four
senses contribute to only 25 percent of the total share. And
man has known this since ancient times. Probably that’s
the reason why the ancient Chinese coined the now popular
proverb, “A picture speaks a thousand words.” It
is very evident that vision is a major source of information
for human beings, and thus if we could possibly provide similar
visual faculties to machines, we shall be able to achieve
visual automation for a very broad range of applications.
Image Processing vs. Computer Graphics
There generally is a bit of confusion in recognising the difference
between the fields of Image Processing and Computer Graphics,
often even in the minds of tech-savvy computer professionals.
Actually, Image Processing and Computer Graphics are entirely
different, almost the opposite of each other. A computer graphics
system is involved with image synthesis, and not recognition
or analysis, as in the case of Image Processing. The input
of a computer graphics system consists of an item list that
describes a scene and its purpose is to transform this list
into a digital image, which could have been formed, if this
scene would really exist. Morphing used in advertisements
could be said to be the most commonly witnessed computer graphics
technique. In contrast, input to an Image Processing system
is always a real image formed via some physical phenomenon
such as scanning, filming, etc. The main role of Image Processing
is not to create information but to extract it, integrate
it, make it explicit and usable.
Applications market
Broadly one can classify the applications areas into four
categories: document and medical imaging, computer vision
& industrial applications, remote sensing & space
applications, and military applications.
From the IT industry viewpoint, Document and Medical imaging
applications are the ones that have proliferated. According
to an estimate, the international market for the first category
alone is growing at about 25 percent CAGR (compounded annual
growth rate), and is projected to be worth more than US$ 30
billion this year. The computer vision and industrial applications
market is still in its nascent stage since common solutions
cannot be applied across a varying range of industrial problems.
The third category of remote sensing and space applications
is mainly in the clutches of government departments or research
organisations the world over, but it is definitely very heavily
funded. NASA, ISRO and NRSA are a few of the organisations
involved in this segment.
The last category of military and classified defence projects
is the one for which the less said the better. Research work
and finances of such projects are kept confidential and are
anybody’s guess.
Coming back to the software perspective, for many years the
computer industry has been promising that document imaging
would solve a stack of paper-related ills. But little came
of those promises, until the Internet began to change the
dynamics of information. In today’s scenario, the benefits
of electronically managing documents are not only surfacing
but are also being widely accepted. Most business information
resides on paper making documents one of the most highly valued
assets of any business. Image processing technology leads
to business applications such as Electronic Document Management
System (EDMS), Workflow, etc, that manage data files and paper
documents alike, to make the “paperless office”
dream a reality. So, document and medical imaging are the
most important applications of image processing technology
as of now.
Office applications
EDMS in its most basic form, is just an archival-retrieval
system. Archive and retrieval document image processing is
concerned with simply replacing paper storage and filing by
electronic movement of records. One could make freehand annotations
or highlight a portion of a document electronically (just
as could be done on paper), along with many other powerful
tools for full-text searching, security and controlled sharing,
etc. Users can achieve huge RoI on this through benefits such
as reduced storage space cost, instant access to documents,
reduced risks of lost or missing documents, and manpower reduction,
thereby providing a cost-effective solution, as also increasing
organisational efficiency.
Workflow solutions automate a business process, in whole or
part, during which documents, information or tasks are passed
from one participant (human or machine) to another for action,
according to a set of procedural rules. Workflow automation
adds the capability to interpret information contained within
the documents rather than just storing them. The solution
is best applied in situations where information is derived
from a variety of sources and there exists a well-established,
multi-stage processing environment.
Forms Processing deals with volumes of paper-based forms for
automatic extraction of data, using technologies such as Optical
Mark Recognition, Optical Character Recognition, Intelligent
Character Recognition, Barcode Recognition, etc.
These solutions find application in several business areas
like banking, financial institutions, telecom, educational
institutions, hospitals, manufacturing, insurance, customer
care, and government.
The Future
Today, we are in the middle of a revolution sparked by rapid
progress in imaging and computer technology. But contrary
to common belief, computers are not as fast as humans in computations
related to the analysis or processing of images. Biological
or human vision is still by far the most powerful and versatile
perceptual mechanism. Scientist Mead notes that “the
visual system of a single human being does more image processing
than the entire world’s supply of supercomputers.”
However, some tasks such as image compression, enhancement
and data extraction from paper via technologies such as OMR,
OCR and ICR, etc, can now be performed on desktop computers
available today.
With increasing power and sophistication of modern computing,
the concept of computation can go beyond the traditional,
sequential Von Neumann architecture and one could even contemplate
optical implementations. A major challenge for automatic image
analysis is that the sheer complexity of the visual task has
been mostly ignored by the current approaches. We (humans)
compute, but not necessarily the way most of the present day
computer systems do. After all, nature did not know anything
about bits and Boolean algebra when vision was created!
It should be noted that the present computer methods, by contrast,
could provide answers only to precisely stated questions that
are not ill-defined. But a ray of hope surely comes from the
distributed and parallel computing paradigms that are expected
to boost real-time response for many image processing solutions.
Image processing technology is waiting to address many unanswered
questions. There is every reason to believe that once this
technology achieves a level of competence that even modestly
approaches that of human vision, and at a competitive cost,
the imaging applications will virtually explode.
Hareish Gur is group head & DGM at Newgen
Software Technologies. He
can be contacted at hareish@newgen.co.in
|
