Thursday, April 25, 2013
INTRODUCTION:
Growth of the Internet and mushrooming of dotcom
companies have made a web designer out of practically everybody. This trend, in
addition, to the easy availability of a wide variety of affordable scanners,
has made flatbed scanners imperative with every desktop machine.
Scanners are no longer
considered expensive, high-end peripherals. They are becoming more affordable
by the day, and also more popular. In fact, with prices on a gradual downward
curve, a flatbed colour scanner is more affordable than even a laser printer.
Traditionally,
design houses have been the prime users of scanners, but the phenomenal growth
of the Internet has made Web designers out of practically everybody. Even the
home user with Internet access wants a home page, photographs and all. All the
more reason to invest in a scanner. Scanners are thus becoming an essential
peripheral for all segments of computer users in India.
Types of scanners:
There are basically three types of scanners-drum,
flatbed and sheetfed scanners.
Drum scanners:
|
These scanners use photo multiplier (PM) tubes and
are expensive, sensitive devices that can capture information at a higher
resolution and higher pixel-depth than
flatbed scanners, which are based on charge coupled devices (CCDs). A photo
multiplier tube is a light-sensiting device with a much higher sensitivity and
lower noise-to-signal-ratio than a CCD. A drum scanner can capture shadow
information that is not visible to the human eye. It can transform that
information into the visible region and improve the image. This is particularly
useful in scanning transparencies. However, drum scanners are too slow and
expensive to use for OCR or document management applications.
Sheetfed scanners:
These scanners take up less desk space and are
easier to install. But most of these can only scan loose sheets. Sheetfed
scanners have been made more compact in size by turning the scanner design
inside out. Instead of the head moving over the paper, tiny rubber wheels,
however, do not track perfectly, so pages at times get skewed, resulting in a
crooked scan. Sheetfed scanners focus more on managing paper flow than on
getting an optimal scan.
Flatbed scanners:
If you have lots of graphics or OCR jobs, a
flatbed scanner is a much better option. With flatbeds, the original page stays
put and scans are much sharper than those obtained with a sheetfed scanner. The
large scanning surface of a flatbed is ideal for odd-sized images. Most flatbeds
also offer optional document feeders that can automatically scan stack of
papers. And they are not as expensive as drum scanners. Some publishing houses
use heavy-duty flatbed scanners.
Handheld scanners:
At the lower end we have handheld scanners or
bar-code readers. These provide some level of portability and functionality at
a low cost. These scanners are typically used to scan originals in strips of
about 4 inches wide and are operated by holding the scanner in your hand and
sliding it over the document. These strips can be reintegrated using special
‘stitching’ software provided with the scanners. However, the quality of
handheld scanners is poor.
Scanning the scanner:
A
flatbed scanner uses a light source, a lens, a charge coupled device (CCD)
array and one or more analog-to-digital converters (ADCs) to collect optical
information about the object to be scanned, and transforms it to a computer
image file. A CCD is a miniature photometer that measures incident light and
converts that measured value to an analog voltage.
When you place an object on a scanner’s copy board
or glass surface and start scanning, the light source illuminates a thin
horizontal strip of the object called a raster line. Thus, when you scan an
image, you scan one line at a time. During the exposure of each raster line,
the scanner carriage (optical imaging elements, which is a network of lenses
and mirrors) is mechanically moved over a small distance using a motor. The
reflected light is captured by the CCD array. Each CCD converts the light to an
analog voltages is then converted to a digital value by an analog-to-digital
converter (ADC), using 8,10 or 12 bits per colour.
The CCDs elements are all in one row, with one
elements are all in one row, with one element for each pixel in a line. If you
have 300 CCD elements for each inch across the scanner, you can have a maximum
potential optical resolution of 300 pixels per inch (ppi) also referred to as
dots per inch (dpi). In case you have 600 CCD elements for each inch, then the
maximum optical resolution will be 600ppi or 600dpi.
There are
two methods by which the incident white light is sensed by the CCD. The first
involves a rapidly rotating light filter that individually filters the red,
green and blue components of the reflected light and is sensed by a single CCD
device. Here, the colour filter is fabricated into the chip directly (as fig).
In the
second method, a prismatic beam splitter first splits the reflected white light
and three individual CCDs sense the red, green and blue light beams. High precision is required in the
optics and in the alignment of the sensing mechanism for this method (as fig).
Resolve the resolution:
A pixel can have many colors (24 bits or as many
as 16.7 million colors), while a dot is a special pixel that has two colors
(black and white). Scanners capture pixels, hence we say pixels per inch (ppi).
Most printers print dots, hence the term dots per inch (dpi). Pixels contain
more information than dots. For scans of photographs you do not need to scan at
the resolution of the printer because the extra information contained in the
pixels is dispersed among the many dots. Even though scanner specifications are
mostly mentioned in dpi these are actually in ppi only.
The true optical resolution is the ability of
scanner to resolve fine details in an original. It is actually the optical
sampling rate or ppi rating of the scanner. The optical sampling rate is the
number of samples captured in the x-axis by the scanner. This is determined by
the width of the scanned area and the number of elements in the CCD. This is
only one of the crucial components of resolution. There can be several other
important factors that determine the final resolution. These include the optics
quality (lens, mirrors, filters), mechanical stability of the optical system,
motion of the carriage, vibration of the object, CCD, and optical components, focal
range and stability of the optical system, impact of temperature and humidity
changes on the optical system, frequency response of the electronic system and
image processing applied to the image.
Thus, a high optical sampling rating
of a scanner does not necessarily guarantee better resolution than one with a
lower optical sampling rate. For example, the resolution of a scanner with a
high sampling rate of 600ppi but low-quality optical system may not be good as
a scanner with a 400ppi sampling rate and a high quality optical system.
Interpolation:
The true optical resolution of the scanner is also
referred to as the horizontal resolution while the y-direction sampling rate is
also called the mechanical resolution or the vertical resolution, since it
indicates the minimum movement of the scanner’s mechanics-the number of steps
per inch that the scanner takes in the y direction. This figure is typically
double the optical resolution. This vertical resolution is interpolated.
Interpolation means the scanner or scanning software generates data based on
the real, captured data. If you see something saying ‘600-by-9,600ppi optical’, this means that it
is a 600ppi scanner, the interpolated number makes no difference whether it is
9,600 or 600, or infinity. Also, in most cases, some marketing masterminds
reverse these specifications. You would be on safer grounds assuming that the
lower number is the actual optical resolution of the scanner.
Interpolation
guesses the values for pixels at a finer level than the scanner samples them,
based on the values of nearby pixels. It is easy to interpolate between two
measurements on the same scan line because the scanner measures the entire line
and has all the information available. It is harder to interpolate in the other
direction –to fill in an interpolated line-because the scanner has not scanned
the lines after the interpolated line yet. By taking extra steps in the y
direction, you eliminate the need to interpolate in that direction.
Scanners that offer higher interpolated
resolutions than the scanner’s optical and mechanical resolutions do their
interpolation for the higher resolutions at the computer. That gives them the
luxury of being able to receive later lines in the image before they
interpolate between lines.
Unfortunately, both the scanner-based and
software-based interpolations can be less sophisticated than the interpolation
routines in a sophisticated program such as Adobe Photoshop. You will often get
better results by scanning at the maximum optical resolution for the scanner,
and then resampling at a higher interpolated resolution in your image editor.
If your work involves scanning photos for Web
pages or output to an inkjet printer, a 300ppi scanner will capture enough
detail for you. But if you are printing to high-resolution output devices or
scanning small targets like slides and enlarging them, or reproducing line art,
then a 600ppi scanner is a much better option.
Scanner software:
The scanner comes with a TWAIN driver, which
functions as a standalone program for scanning, and enables you to scan images
directly into most Windows programs. If an application supports TWAIN, it
usually lets you call up the TWAIN driver by entering a scan command. If it
supports OLE instead of TWAIN, you can use OLE to call up the driver and insert
the scanned image. In addition, all scanners today also come with image editing,
OCR and sometimes other applications as well.
Image quality:
In the ultimate analysis, the most important issue
for any scanner is image quality. Understanding what affects image quality will
not only help you make a more informed buying decision but will also help you
take best advantage of whatever scanner
you get.
High on the list is colour balance-the ability to
capture neutral colors in neutral form. If you scan a black-and-white photo in
colour and the result has a colour tinge, you can be sure that the colors will
be off on your colour photos too. A closely related issue is colour accuracy,
or the ability to capture colors that closely match the original. Many scanners
often impart a pinkish tinge to scans.
For any given scanner at any given setting, both
tonal quality and the colour accuracy will vary depending on the screen or
printer you are using to look at the result. That is why an easy-to-use
calibration feature is critical. You must keep in mind that images that look
good on screen will not necessarily look good when printed, and images that
look good while printed may not look good on screen.
You will want to
calibrate for both screen and printer; making sure you use the appropriate
calibration file when you scan, typically by telling the TWAIN driver the final
destination for the image.
Another important image quality criterion is gamma
correction. Gamma correction essentially lets you modify contrast level at
different levels of brightness. Changing the gamma setting can make a
tremendous difference to an image.
Noise level
and the closely related signal-to-noise ratio is the ratio of the usable
signal, in this case the image to noise in a scan. Noise is a distortion in the
image’s analog signal. This is an analog problem and is confined to the analog
electronics in a scanner. You can change the signal-to-noise level
substantially by changing settings in the TWAIN driver. Turning down the
brightness settings also reduces the amount of noise in an image.
Colour registration-a measure of how well three
colors line up with each other-is not usually an issue for colour photos but
may be an issue if you are scanning line art. If the registration is off, you
will see an extra ‘halo’ of colour at the edges. When you are scanning to a Web
page, even a full pixel off is generally acceptable in a 600-dpi scan, though
it will be noticeable to those who look for it in 300-dpiscan.
Scan quality:
The most difficult areas for the scanner interpret
correctly are the very dark and bright areas of a picture.
To determine the scanner performance, a still life
photographs with areas that were observed for correct interpretation by the
scanner. The image was also checked for any signs of blurring and cloudiness or
whether the dark areas near the leaves and lamp were properly distinguishable.
Finally, the colors of the image were checked if they matched those in the
original photo. The image output for each scanner is shown next to the individual
review. In some cases, the images have been zoomed to show problem areas.
For
determining the ability of the scanner to differentiate between dark, bright
and colored areas, we used an IT8 card. This sheet was scanned and then
compared with the original using Adobe PhotoShop 5.5. Using the histogram
function, we determined the amount by which the scanner could differentiate
between the light strips at the top of the card, the monochrome areas at the
base and the dark area at the top right of the card.
Features:
We also awarded points to the overall ease of
setup and installation of the scanner. It included the ease of connecting the
device and installing the driver and other software, whether the scanner was
sturdily built and had a firm hinge for accommodating thick documents like
books and reference cards. We also checked for features like an integrated
power supply, a transport lock, and for any noise produced while scanning. One
of our criteria was also the comprehensiveness of the manual (electronic or
printed) supplied with the scanner.
The bit mystery:
How can you identify a first-rate scanner? The
answer isn’t so simple, though it would suffice to see that it is one that
combines accuracy and minimal loss with maximal power and ease in compensating
for the distortions. One of the most potentially confusing claims for the
scanners is the number of bits that they offer. You often come across terms
like 24,30, or 36 bit scanners, but what do they actually mean?
Bits are the basic elements of
digital data. A single bit is either on or off, usually expressed as 1 or 0 so
that there are only two variations. Each pixel of a scanned image has a depth
of one to 32 bits. 1-bit images are black and white (for example, line art). A
2-bit pixel contains four variations (00 01 10 11) and allows a variation of
color from white to light gray to dark
gray to black.
A 8-bit pixel can vary anywhere within the
full range of 256 gray values. 24-bit images are actually three 8-bit channels,
one each for red, green, and blue light. A 32-bit image can be an RGB image
with a fourth channel (for example, an alpha channel in Adobe Photoshop) or,
more commonly, a CMYK image with one 8-bit channel for each of cyan, magenta,
yellow and black.
A 24-bit scanner divides each of
its primary colors-red, green, or blue (RGB) into 8 bits, or 256(28) shades.
(In 256 shades of gray). This 256 gives you 16.7 million possible colors. This
is the maximum that most high-end
graphics boards will display. A 30-bit scanner offers 10 bits or 1,204 shades
for each color, while a 36-bit scanner offers 12 bits or 4,096 shades.
However,
even this data is to enough for a perfect picture. Once the seventh and eight
bits are reached, accuracy takes a nosedive. This results in loss or distorted
details-especially in the highlight and shadow regions. Further, applying tonal
corrections-adjusting gamma curve results, brightness and contrast-reduces the
size of the color palette, resulting in loss of data. You can still change the
colors but you cannot work with what is not there!
The solution lies in adding more bits and making it a
30-bit scanner. A 30-bit scanner offers 10 bits per color or captures 1,204
shades of red, green, or blue for over a billion colors. The first 8 bits are
fairly accurate, so the scanner can just junk the last couple of bits ad leave
you with superior 24-bit color.
If the scanner performs
tonal corrections of color at the hardware level greater advantage can be
derived from the larger number of colors. If your expand part of the tonal
range of a 24-bit image to bring out, say, the shadow details, you
automatically end up compressing it else-where thus losing midtones and
highlight details. A 30-bit scanner can use your instructions to select and
deliver the best 24 bits of data for your needs, giving you the corrected image
with much greater detail retention and smoother tonal continuity.
However, 30-bit and 36-bit scanners use the extra bits
internally, and generally send only 24 bits to the computer.
COLOR MODELS:
Color models are closely related to bit depth. Grayscale
goes up to 8-bit, which renders 256 shades. Color images are multiples of 8-bit
channels. RGB, the normal model for computer graphics, goes up to 24-bit (three
8-bit channels for red, green and blue). CMYK, the standard for printing color
images, is a 32-bit model.
RGB color:
Red,
green and blue are the primary colors of light. The human eye responds to
stimuli from varying RGB wavelengths and renders the appropriate signals to the
brain so that we perceive such colors as cherry, mauve and beige. Most scanners
use an RGB color model for recording digital image data. RGB color is called
additive because colors throughout the spectrum are created by adding varying
intensities of red, green and blue light to black (no light). These intensities
vary from 255 (full intensity) to 0. Each color channel 256 variations and
their combinations allow creating a total of 16,777,216 colors. A combination
of R: 255 G: 255 B: 255 creates white, while R:0 G:0 B:0 is black (no light).
CMYK COLOR:
Cyan, magenta and yellow are the secondary colors of RGB
and are opposites. When RGB light strikes an object, the amount of cyan,
magenta and yellow in the object’s pigmentation affects how much light is
reflected back (It’s the reflected light that we se). cyan absorbs red light,
magenta absorbs green light and yellow absorbs blue light. The degree of
absorption depends on the amount of pigment or, in printing terms, the amount
of CMY ink. This is why CMYK is considered subtractive; the colors displayed by
CMYK are the result of subtracting varying amounts or red, green and blue
light.
CONTACT IMAGE SENSOR (CIS)
TECHNOLOGY:
Contact
Image Sensor (CIS) is a relatively new sensor technology for flatbed scanners.
CIS scanners deploy dense banks of red, green and blue LED’s to produce white
light and replace the mirrors and lenses of a CCD scanner with a single row of
sensors placed extremely close to the source image.
A CIS scanner uses a single chip that handles many
data-processing functions. It is made of silicon, and use on-chip filters to
separate light into red, green and blue primaries. But while a CCD requires a
separate component, an analog-to- digital converted (ADC), to translate the
sensor data into binary information, a CIS sensor has onboard logic to perform
the task of the converted.
CIS-based scanners requires fewer supplementary components
than CCDs, as a result they require less space, less power and cost less to
manufacture. But current CIS sensors tend to have a smaller dynamic range than
CCDs. In addition, because CIS scanners are relatively new, manufactures don’t
have as much experience at fine-tuning the noise reduction and filtering
algorithms as they do for CCD scanners.
CONCLUSION:
Finally I would like to conclude that we
do have a problem with scanners of matching the exact colors. Suppose take a
target and scan it, view it on the screen or print it and compare the result
with the original copy. It would, of course, be wrong.
Why because all scanners and monitors
define colors with device dependent color models, so the definition for any
given color varies from one device to another.
Finally, what I mean to put
forth is to find the ideal solution to judge the scanners in the context of the
standard color space or color management system that all the devices on the
computer-scanners, monitors and printers-are actually trying to match.
