Laureate Online Education Internet and Multimedia Technolog.docx

Laureate Online Education Internet and Multimedia
Technology
© All right reserved, 2000-2008. The Internet and Multimedia
Technology module, in all its
parts: syllabus, guidelines, lectures, discussion questions,
technical notes, images, labs,
projects and any additional material is copyrighted by Laureate
Online Education B.V.
Module: MSC MD (INTMMT) Seminar 3
Bitmaps and Colour
Introduction
We will turn our attention to Bitmaps and Colour during this
Seminar. As we did
in Seminar 2, we will again recommend some applications and
encourage you to
experiment.
Bitmapped images have none of the complexity of 3D graphics
and yet they are
the final output from a 3D system. Often the 2D image desired
will first be
created in 3D and then, after the perspective transformation is
applied and it
becomes 2D, modified and enhanced. Let us take a look at some
ideas
associated with Bitmaps.

We start with the resolution or measure of the granularity of the
image. It is a
common error to assume that an image having 640 by 480 pixels
on the screen
cover s only that number of phosphor dot triads on the face of
the monitor. A
screen may state it provides so many dots per inch as its
resolution. This is a
measure of the manufacturing quality. Depending on the size of
the screen we
will have a different number of dots per pixel for a given
computer resolution. It is
also rather short sighted to keep an image at a small resolution
since it will look
very ‘blocky’ if enlarged or viewed at a greater resolution than
originally stored.
This gives rise to large files being kept just in case we need the
resolution! One
obvious way around this is to compress the image.
Compression schemes are considered to be of two types; Lossy
and Lossless.
With lossless compression we can always regain the original
quality whereas
with lossy compression we sacrifice some aspect of quality
depending on the
application. When dealing with multimedia and the Internet we
are very
concerned to obtain the best compression ration for a given
quality of service
(QoS). The compression ratio is given by;
Compression Ratio = Uncompressed file size
Compressed file size

Most compression schemes rely on taking advantage of
redundancy in the data
(in our case, the stream of pixels which make up the image).
This takes many
forms. If we examine one scan line of an image we will notice
that many pixels
1
are the same colour so stating for example Red 100 times is
wasteful. It is
obviously shorter to store the value for red and its length. This
is the basic idea
behind Run Length Encoding (RLE). A simple compression
related to colour is to
appreciate that we do not need the full range twenty four bits
can provide. If we
sample the image for the 256 commonest shades then we can
reduce the
storage for colour by two thirds by sacrificing fine shading
differences. A further
approach is useful with video and makes use of the fact that
there is redundancy
between frames. Not all components of the image move from
frame to frame so
we need transfer only the changed elements. This is part of the
Motion Picture
Expert Group (MPEG) standard. We will look at that when we
discuss video in
seminar 4.
The most common compression standard used in multimedia
work today must be
the JPEG system from the Joint Photographic Expert Group

(JPEG). JPEG is a
group of experts (currently over 200 worldwide) who are
nominated through their
individual national standards organisations. It may surprise you
to know that this
group started work in the late 1970’s on video text! They are
currently working on
JPEG 2000. Several new features are being implemented
particularly several
which aid e-Commerce. E.g. Encryption, watermarking,
licensing and
registration. These will match similar work on MPEG 7 and 21.
These new
standards although not yet common will change multimedia
production and the
way we pay for it. JPEG 2000 is available as a Photoshop plug-
in and many
companies are now making it available, such as; Sanyo on a
digital camera,
QuickTime and Acrobat Reader version 6. Look for its file
extension .JP2.
The current JPEG standard uses the Discrete Cosine Transform
(DCT) applied
to 8x8 blocks of pixels taken from the image as the basis for
compression. See
figure 1.
8x8
image
DCT

Transform
Coefficient
Quantization
Lossless
Compression
stage
Figure 1. A block diagram of a JPEG encoder.
The DCT produces a set of frequency responses representative
of the contents
of the series of blocks in the image. For each of these
frequencies the values of
the coefficients are stored. The coefficients produced by the
DCT are quantized
and those below a certain value are discarded. In more detail;
the difference
between the coefficient values and a given level provided by a
table is divided by
a constant and then stored. The stored values below a threshold
are then
2
discarded. These are values which represent high frequency
terms and therefore

very fine detail and facilities are available to tailor this area.
Finally the data
stream is Huffman encoded so that no further data is lost.
Huffman coding is
used because it is a very efficient lossless method. Remember
JPEG is lossy
due to quantization but this final stage is not.
We will end this lecture with a brief look at some manipulation
techniques on
Bitmapped images. Many of these techniques have their origins
in the Image
Processing literature. We cannot cover that field in this module
so you are
encouraged to look at any standard reference work on the
subject or for
particular topics use the web. (For a selection, see (11 In the
lecture of Seminar
2)).
Since bitmaps are collections of pixels we cannot work on a
discrete area directly
so we are forced to consider individual pixels. We rapidly
realise that is not an
option! We need operations that let us work on collections of
pixels
simultaneously. One class of operations affect the whole of the
image, these are
masking techniques and are used for smoothing, noise
suppression or special
effects. They take advantage of a mathematical technique
known as convolution
and use a convolution mask to scan the image.
Referring to Figure 2 which shows an example based on a 3x3
pixel mask used
as the convolution kernel, we see the new image being

developed bit by bit. The
mask is moved across the image and actions are taken on the
image pixel value
in the centre of the mask based on the values in the mask. These
produce a new
modified image in the buffer. So much for the process but how
does it work? Why
don’t we look at one operation in particular. Have a look at
figure 3. The figure
shows a small image (only nine pixels) which we wish to
enlarge. We need to
find some new values to go where the zeroes have appeared in
the larger image.
Buffer
Shift and sum
until end of row
Result Mask
Image
Figure 2. The convolution mask generates a new image as it
moves.
3
Figure 3. A small picture of nine pixels was enlarged on the
right but it now has zero values in the
new pixels. We can fill them by averaging surrounding pixels.

Figure 4 shows a mask which will produce the desired result.
Try it for yourself.
1 12 21
3 45 23
4 22 12
0 0 0 0 0 0 0
0 1 0 12 0 21 0
0 0 0 0 0 0 0
0 3 0 45 0 23 0
0 0 0 0 0 0 0
0 4 0 22 0 12 0
0 0 0 0 0 0 0
0.25 0.5 0.25
0.5 1.0 0.5
0.25 0.5 0.25
Figure 4a. The mask we will use to put average values in place
of the zeroes in
the enlarged image.
0.25(0) + 0.5(0) + 0.25(0) + 0.5(0) + 1(1)+ 0.5(0) + 0.25(0) +
0.5(0)
+ 0.25(0) = 1
Shift right and repeat
0.25(0) + 0.5(0) + 0.25(0) + 0.5(1) + 1(0)+ 0.5(12) + 0.25(0) +
0.5(0)
+ 0.25(0) = 6.5
Note: Original values remain the same

Figure 4b. This is the upper left quadrant of the enlargement.
4
We start in the upper left quadrant (the yellow and green
square) and centre the
mask on the image value in the centre of the mask then we move
to the green
and blue square, and continue on in this manner.
Figure 4b shows how we add each result of an individual
multiplication on the
overlaid pixel. Notice that the operation has no affect on the
original value but
puts local averages in place of the zeroes.
Different values of the convolution mask will perform, as
explained in the
textbook, different effects.
To get acquainted with the Bitmap programs see the Tools
section at the end of
this lecture.
Colour
We learn about the basics of colour in our early physics classes
but there is
much more to colour than the discussion of its component parts.
It is colours that
confirm our assessment of the appearance of an object. If we are
to successfully
replicate objects in our websites or Multimedia productions we
should understand

how colour interacts with objects and how we interpret that
interaction. As a
production team member we must ensure consistency across our
product and
that means across the colours used. We are going to examine
these issues in
this part of the seminar.
The assessment of appearance is not only influenced by the
geometry and
colour of the region but also by our expectations. We deal with
images
presented to us by an opto-electronic system, however given the
same sample;
different systems will produce visually different images. This
has many causes;
from the variations in system components to the variability of
viewing conditions.
In the case of colour every observer will describe a colour based
on their own
experience of colour naming and their ability to discriminate
colours. A simple
means of ensuring that the observers are considering the same
area is for all
parties to be present and observing the same image. In a modern
Multimedia
development team, this is not practical; therefore, some time
will elapse between
consideration of the image by different people. In this case it is
possible for the
subsequent observer to fail to notice (or be incapable of
noticing) a colour
change or to misinterpret the extent of the particular colour
changes under
discussion. If a product should require further quality
assurance examination

then the differences in images may be sufficient to deny a
successful
assessment. The evaluation method chosen must therefore be
precise and
reproducible. This mandates a standardised set of conditions
should be
implemented. Particularly important aspects are lightsource
spectral content,
intensity, direction and also the viewpoint plus the imaging
equipment. Some of
these items, such as lighting direction and viewpoint are
currently not controllable
in 2D packages but the light intensity, spectral content, imaging
and ambient
viewing conditions are.
5
We are concerned with the reproducibility of colours with
consistency and fidelity
at the developer's monitor and the monitors of subsequent
viewers after the
image capture and processing stages. This will ensure
repeatability when
measurements are made. These concerns are dealt with by the
appropriate
Committee International d'Eclairage (CIE) standard. (The CIE
is the recognised
international standards body for colour.) A secondary concern
is the effect of
viewing conditions upon the perceived images. These are
examined in this
seminar and recommendations are made to aid in consistency of
colour

assessment.
We know that colour is a sensation produced because certain
electromagnetic
frequencies have a physiological effect, but what do we
understand by
appearance? For example, we talk of objects being 'glossy'
without considering
the precise nature of glossiness. This section will help define
some descriptive
terms and how they arise as object attributes. Appearance is
determined not
only by the geometry, colour and surroundings of an object but
also by our
subjective understanding of the nature of the object. Certain
clues are missing or
modified because we are always dealing with an image and not
the object, i.e.
Depth cues are distorted if we use wide angled lenses. The
appearance of an
object can be described by classifying them according to their
effect on incident
light. There are two further sub-divisions we can use to clarify
this interaction.
These are termed the colour and geometric attributes. Prior to
looking at this
classification, it is necessary to introduce some terminology.
The amount of specular reflection depends upon surface
smoothness, the
refractive index of the material, and the incident angle of the
light. This produces
what we term as glossiness. Non-metallic materials make little
change to this
reflected light so that a highlight is formed which is the same
colour as the
lightsource. If the surface is rough then this light is scattered

and is added to the
diffuse light. Thus, the diffuse light is modified by the
lightsource colour, i.e. a
white lightsource will cause the object to appear lighter in
colour. Most of the
light is refracted into the material where it encounters the
pigment particles.
Multiple reflections and refractions occur; thoroughly diffusing
the light until it
finally leaves the surface in all directions. This is known as
diffusely reflected
light. It is the refraction of the light into, and transmission
from, the pigments that
produces the colour. The spectral content of the light passing
through the
pigment is selectively absorbed thus producing colour. For
example, a green
pigment selectively absorbs blue and red.
Metal surfaces have negligible penetration by light and almost
total reflection
occurs producing their shiny appearance. Colour in metal is due
to certain
wavelengths being more efficiently reflected than others, e.g.
Brass absorbs
some blue wavelengths hence it has a yellowish appearance.
Translucent objects allow more light to pass through or to be
transmitted than
opaque materials hence their appearance owes more to diffuse
transmission.
6

Finally, transparent objects allow almost all the incident light to
be transmitted
through the object.
Now we are in a position to characterise our objects:
a) Diffusing surfaces are those that possess surface colour.
Their colour
attributes come from diffuse reflection and their geometric
attribute is glossiness
from specular reflection. Their colour attributes are described
in terms of
lightness, hue and saturation.
b) Metallic Surfaces possess a colour attribute from specular
reflection. They
are described by their glossiness, hue and saturation. Their
geometric attribute
is a slight colour blurring due to surface diffusion.
c) Translucent objects have colour mainly from diffuse
transmission. They
possess colour attributes of translucency, hue and saturation and
geometrically
glossiness by specular reflection.
d) Transparent objects have colour from transmission. This is
described in terms
of clarity, hue and saturation. Geometrically there may be a
slight blurring due to
diffuse transmission.
The following paragraphs explain how we can maintain colour
consistency
across systems and how we may assess a given colour against
another. This

would be very important for example, if you were producing a
multimedia product
for a fashion house.
In order to assess colour systematically and consistently, it is
necessary to
quantitatively evaluate the light from the surface or object.
Quantitative
evaluation of light from objects is particularly important to
manufacturing
industries involved in presentation products. i.e. washing
powders, paints, and
much of the research is directed towards their needs. Test
samples are usually
made of flat uniform materials and the light is controlled to
ensure stable test
conditions. The light may be reflected or transmitted in a
different manner for a
given material depending upon the following:
a) The incident angle of the light.
b) The observer's viewpoint.
c) The spectral content of the lightsource.
d) The refractive indices of the materials constituents.
Two instruments are used to measure light leaving an object.
The
spectrophotometer measures light in terms of its constituent
frequencies. The
goniophotometer measures the quantity of light emitted in a
particular direction.
Both of these devices are used to produce data for comparison
against known
samples characteristics. When collecting this data it is
necessary to control all
the above parameters, a) to d), plus the following:

7
e) The intensity of the light source.
f) The angular size of the light source.
g) The spectral response functions of the observer.
Colour evaluations are made using a diffuse source with diffuse
geometry whilst
for gloss it requires a well-defined (possibly patterned ) source.
From this data we can develop models of how materials modify
light for inclusion
in imaging light models. We can examine the items (a-g) above
to see which it is
possible to standardise or hold constant and from this develop a
best practice. In
photographic and film studio light sources, the intensity and
spectral content are
both controllable and controlled so that by recording their
values at the time an
image is taken we can provide transformations for other
conditions. The angular
size of the lightsource may be fixed or diffuse. Correction can
be made to an
image for poor light distribution provided the light type and
value is recorded.
The viewpoint of the observer is of concern when comparing
samples e.g. for
historical purposes, but is in practice uncontrollable. Computer
processing can
be enlisted in certain circumstances to change the viewpoint
over small ranges.

The major characteristics of human vision are well known and
can be found in
any introductory text but their effect on fidelity of reproduction
may not be
obvious. This section looks at factors affecting our ability to
assess the accuracy
of the image both on the monitor and through any attached
computer system.
The human eye can readily discriminate between 10 million
juxtaposed surface
colours, however, from memory, we can only manage to recall
approximately
300. It is therefore, of primary importance that our image and
the original are
viewed together and as close to each other as possible if fidelity
is to be
achieved for a given sample. This is evidently not feasible. We
must then resort
to indirect means for colour and texture matching such as
comparison against
standard samples. In order to do this multiple monitors must be
calibrated
against the samples from time to time. A simple approach is to
point the camera
at a known sample, freeze the image and then compare the
sample to the image.
This is subject to gross errors that are covered in the text. An
analytic approach
is developed later.
The human eye functions most efficiently in daylight. Thus for
maximum
precision we should view our image and object at light levels
consistent with
daylight. This is possible for object-object comparisons but

difficult for images
due to the technology used. An equivalent approach is to
ensure the spectral
content of the lightsource and our viewing conditions are
matched. With our light
emitting screen technologies based on three chromaticities this
implies matching
the external light source to them and the monitors to each other.
Background distractions add coloured light to a scene and the
recommendation
is that they are removed and replaced with a neutral grey. This
also helps with
8
chromatic induction and chromatic adaptation. Chromatic
induction is the
process whereby colour apparently changes due to preceding or
surrounding
visual stimulation. It has physiological causes due to gain
changes in
photoreceptors and polarisation of chromatic sites. The induced
hue is
approximately complementary to the inducing field. Chromatic
adaptation is the
ability of the eye to adapt to colour changes due to changes in
illuminant. This
leads to the concept of corresponding colours; sets of
tristimulus values that
describe the stimuli that evoke the same colour appearance
under different
conditions, eg. different light sources. The implication here is
that we should
control the ambient lighting conditions when viewing the

monitor or a photograph.
We should discuss the lack of correlation between
instrumentation and our
eyesight. The tristimulus theory of colour reproduction has
served well as an
enabling methodology for the construction of the commonest
imaging
technologies however it has long been known that our vision
sense is influenced
by many other processes. The concept of colour opponent cells
was first hinted
at by Hering in 1878 and later developed by Mueller in 1930.
This provides a
more complete theory and explains our ability to notice small
changes in hue at
high saturation levels whilst requiring a greater relative change
in saturation for a
particular hue before it is noticed. Thus, it is noted that our
eyes are not linear in
their response to different stimuli. The tristimulus system
represented by the
RGB colour space has a linear basis so that other scales are
required to deal
effectively with colour differences. Typical of these are the
CIELUV and CIELAB
scales Our first difficulty; when describing the relationships
between these scales
mathematically is that we can only use approximating formulae.
Implicitly
instruments may not be able to produce (or describe) the colour
required. The
eye responds to factors external to our sample region of interest
so that a
measuring instrument may offer an accurate assessment based
on recognised

standards but we would still dispute their validity. Textured
objects provide a
further problem by distracting us from a clear colour
assessment. They may also
confuse instruments in a different sense thus producing larger
apparent errors.
The final point we must remember is that human perception is
variable. Scales
originate as collections of data from thousands of individuals
and are not
definitive of an individuals colour perception.
The monitor is easily the greatest contributor to variability of
colour output. The
ideal monitor would exhibit the following characteristics:
a) Phosphor Constancy. The spectral output would not be
affected by the
voltage applied.
b) Phosphor Independence. The output of a phosphor would
not be affected by
the output of other phosphors.
c) Spatial independence. The output of a given location would
not be affected
by the output at other locations.
d) Uniformity over time. The characteristics would not vary
over time.
9
e) Spatial Uniformity. Output characteristics would be
identical across the
monitor face.

All of the above cannot be totally controlled which causes
variability in images
across monitors. There is little that can be done outside of
improved quality to
change items a)-d), Lack of spatial uniformity can be
compensated for under
program control. The calibration of monitors is necessitated by
their temporal
instability and is reviewed later.
Any monitor has a specific range of colours which are
determined by the
chromaticities of the CRT phosphors thus certain visual
chromas are outside the
realizable range. When different monitors are used to display
the same image
they will require matching to the original monitor. For an
attached computer
system the monitor differences are further complicated by the
image capture
system.
The device which isolates and captures an image for computer
display is known
as a scanner for 2D work and a frame grabber for video. This
converts the
incoming analogue signals to a digital format and is therefore
limited in spectral
bandwidth by quantization noise.
We will develop a method of matching the various system
components
throughout the remainder of this section. To begin with it is
necessary to have
some standard to which we can refer our subsequent analysis.
The next section

provides the relevant background.
Monitors are provided by different manufacturers each of whom
strive to offer a
superior specification. This may often be nothing more than an
alternate
specification. However all manufacturers provide equipment
built to international
standards and by using them as our baseline common ground
may be
established. The CIE have produced standard chromaticity
diagrams against
which various colour systems can be measured. The standard
for monitors is the
CIE 1931 2o Standard Observer chromaticity diagram.
The CIE standard chromaticity charts have several inadequacies.
It should be
noted that they are not a good choice for object descriptions
since they refer to
combinations of illuminant and object, nor do the CIE
dimensions relate to
perceived attributes of colours such as intensity. The CIE
scales are linear and
hence not uniform when visual spacing of colours is
considered.. For an applet
which shows the diagram see (Schaller, n.d.)
In the search for consistency and fidelity it is necessary to
understand how an
image may be degraded before we can determine corrective
measures.
An important quantitative measure of performance in video
equipment is the
signal to noise ratio (s/n) ratio. This is often used as a measure
of performance

for analogue equipment. Noise is an unwanted, often random
addition to the
10
wanted signal level. For professional colour cameras a typical
figure is 46 db's,
or approximately a 3% disturbance to the wanted signal level.
This type of
degradation is cumulative throughout the analogue system
which implies that at
each stage in the system the input quality is fixed and any
losses due to the
previous stage become unrecoverable.
The process of degradation is described by taking each
component of our
system in turn.
In the camera the separate RGB signals are converted to
composite video
according to the local standard format i.e. NTSC, PAL,
SECAM etc. There is a
consequent reduction of chrominance bandwidth (NTSC being
the worse)
because of the historical necessity to provide compatibility with
monochrome
systems and international radio frequency bandwidth limitations
for television
channels from which the composite definitions were derived.
Issues such as radio frequency (RF) interference are considered
negligible in our
localised system since there are no broadcast requirements.

Unfortunately every
connection can lead to colour balance problems due to induced
differences in
electrical path length. The loss of high frequencies and hence
part of the
chrominance signal is also noticeable with increasing path
lengths due to
capacitive effects.
The video recorder losses are mostly due to the magnetic media.
Magnetic
media is good for digital storage but not for analogue storage
due to its inherent
non-linearity. For this reason the signal is modulated onto a fm
carrier prior to
storage. VHS, beta, and 8 mm formats all use a component
system which
separates the signal inside the recorder into luminance and
chrominance
channels. On replay the signals are recombined thus two stages
are susceptible
to losses in bandwidth and increased noise. Small tape size and
low transport
speed are further reasons for poor bandwidth.
A frame grabber is a source of sampling errors. No matter what
number of bits
per colour per pixel is chosen there will always be some
quantisation error. This
can increase artifacts produced by the analogue system. The
simple solution is
to capture images at the highest possible bits per pixel per
colour available.
Images can be captures from digital cameras, screen captures
and scanning of

previously captured images (on film and books). For a tutorial
on how to scan,
which is an art by its own right, see (Fulton, n.d.) For a tutorial
(one of many) on
the use of digital cameras see (Curtin 2007).
The computer has the inherent ability to store information
accurately and
indefinitely. This allows us to say that the CPU is not a source
of degradation
however software file manipulations can be if the data is
transformed between
11
file formats into a different number of bits per colour.
Therefore, it is necessary
to store the data at the resolution of the frame-grabbed image.
The display system was originally designed for motion
broadcast and
compatibility with monochrome technology. This effectively
fixed the channel
bandwidth and the line rates for display on television monitors.
Modern computer
monitors have higher definition screens with associated
electronics capable of
greater fidelity if properly adjusted. The colour output is based
on the phosphor
chromaticities for a particular manufacturer’s product. The
monitors with greater
colour range will have the widest spacing between their
phosphor chromaticity
values. Differences in these values between monitors causes

them to produce
different colours for the same input data if matching is not
carried out.
The maintenance of fidelity and consistency within the chosen
system is beyond
the operators control except for insisting on regular
maintenance checks.
Regular checks should include a test chart to ensure correct
focus is maintained
and calibration of the lightsource intensity controls. Regular
calibration must also
be carried out on the monitors at least every six months. This
ensures the
monitor has known characteristics. Importantly, a known white
point must be
established which can then be used in determining
transformation matrices
between one monitor and another. This is done using a
colorimeter, which
compares a known white spot of light to the monitors white.
These transformation
matrices are the key to consistent reproduction. Given two
monitors with known
phosphor chromaticities (these can be measured but are usually
available from
the manufacturer), image data can be mapped from one system
to the other so
that the colours produced remain the same. This is achieved by
mapping the
monitors RGB space to the CIE space and then back into the
other monitors
RGB space.
The old standard for computerised image storage was eight bits

per colour per
pixel. This is true for the internet today. This is not sufficient
for larger areas
requiring subtle shading i.e. skin and flesh tones in close-ups,
where artefacts
such as mach banding become apparent. Quantization would
ideally take place
at 10 bits per primary to avoid our detecting such effects. There
is a need for the
associated computer graphics card to handle at least 24 bits per
pixel if we are to
maximise fidelity and consistency.
Image processing software requires an accurate lighting model
with the attendant
cost in computational time. Most common models do not
adequately consider
the materials characteristics. They take the lightsource as the
primary factor in
determining the resulting colour at a pixel and calculate
contributions for three
frequencies only i.e. RGB. For greater accuracy, we must more
fully use the
spectral curve data for both the lightsource and the material. If
we are to
produce accurate models of real world objects then this
computational cost must
be accepted.
12
The maintenance of fidelity from the multimedia system
viewpoint is easily

summarized; you get what you pay for. There is no substitute
for quality.
Certain problems cannot be overcome without the aid of image
processing
techniques. These may be incorporated in a package such as
Photoshop. It is
not sufficient to simply store the image, due consideration must
be given to the
format used. In particular, the format must also be capable of
storing the monitor
transformation matrix, the light source intensity and general
system parameters.
The TIFF standard is a representative example of a file format
with the necessary
capabilities as are the more common MPEG and JPEG
standards.
We have already discussed the effects of chromatic induction
and adaptation that
help make the case for consistent viewing conditions. The
modern multimedia
system gives scant regard to the placement of the monitor
beyond ease of
viewing yet the ambient light affects the quality of the
perceived image. The
colour of an image on a monitor is diluted due to the specular
diffusion of
ambient light at the screen.. At higher light intensities, the
screen appears white
and impossible to use due to what is termed glare. This is a
function of the
viewers position with respect to the light sources. Several steps
can be taken to
minimise these effects. Firstly, the ambient light should be kept
as low as

possible consistent with working requirements. Secondly, the
room should not
admit light from external sources and ceiling lights should have
parallel diffusers
fitted. The spectral content of the lights should be as close to
daylight as
possible.
The manner in which the monitor is used can affect colour
quality. Many high
quality monitors use a CRT that contains a shadow mask. Due
to thermal
excitation by the electron beam the mask will expand or 'dome',
this will cause
colour shifts. It is therefore necessary that the monitor is
switched on at least
half an hour before use if colour comparisons are to be made.
Finally, any
monitor has greatest chromaticity range at low beam currents in
the CRT. The
brightness control should therefore be set as low as possible.
The consistency and fidelity of colour reproduction across
systems can be
achieved by adopting as a set of guidelines the
recommendations in the
preceding sections. These are listed below:
a) View images under standard conditions.
b) Ensure the monitors are calibrated regularly.
c) Switch on the monitor half an hour before use.
d) Store images in a flexible format such as TIFF.
e) Include with the image data; the light intensity, the matrix
conversions, and

the system data.
13
Certain problems are currently inescapable. These are;
the analogue front end, (Rapidly being resolved)
the CCD lens pixelation,
the display phosphors technology,
the age of the viewer.
Finally some recommendations for a cost effective system for
colour analysis of
images are offered. These are aimed at what can be reasonably
achieved
outside a professional image processing studio:
a) Use S-VHS or direct RGB (preferable) to the frame grabber
b) Sample at 24 bits or greater
c) Use a 24 bit frame buffer and do not re-sample for storage.
d) Display using the frame grabber definition
e) Purchase only good quality monitors and have them
regularly
calibrated.
The need for consistency and fidelity is greatest in picture
analysis (i.e. colour
edge detection), pattern matching, and picture archiving
applications. This
section has examined some of the problems and issues involved
in reproducing
accurate images and descriptions of surfaces. The means to

attain consistency
and fidelity have been discussed and some simple suggestions
have been made
with regard to best practice. In particular, the use of digital
techniques where
possible and the need for the highest sampling rates available
are stressed.
We have come to the end of our seminar on bitmaps and colour.
It has covered
some very complex topics in as straightforward an approach as
possible.
Links and References
Curtin D.P. (2007) A Short Course in using your Digital Camera
[Internet]
http://www.shortcourses.com/using/index.htm
(Accessed: 15th September 2008)
Fulton W. (n.d.) A few scanning tips [Internet]
http://www.scantips.com/
Schaller N.C.(n.d.) A CEI chromaticity applet [Internet]
http://www.cs.rit.edu/~ncs/color/a_chroma.html

14
http://www.shortcourses.com/using/index.htm
http://www.scantips.com/
http://www.cs.rit.edu/%7Encs/color/a_chroma.html
Tools:
Bitmap 2D:
Commercial: Photoshop, Photoshop Elements, Print Shop
Pro
Bundles: Many. Simple editors are available with almost
every digital
camera.
Simple editors are found also in the Windows system (Paint
and sometimes
Photoed (more advanced))
Freeware:
Gimp (very advanced!): http://www.gimp.org/
Irfan, A very good viewer (not a manipulator):
http://www.irfanview.com/
Colour:
Freeware: Pixie, a small program that analyzes colours:
http://www.nattyware.com/pixie.html
The CEI chromaticity applet, mentioned in the
links, can also be
downloaded as a program from the same site:
http://www.cs.rit.edu/~ncs/color/a_chroma.html
15

http://www.gimp.org/
http://www.nattyware.com/pixie.html
http://www.cs.rit.edu/%7Encs/color/a_chroma.htmlBitmaps and
Colour

Laureate Online Education Internet and Multimedia Technolog.docx

More Related Content

Similar to Laureate Online Education Internet and Multimedia Technolog.docx (20)

More from DIPESH30 (20)

Recently uploaded (20)

Laureate Online Education Internet and Multimedia Technolog.docx