Professional Documents
Culture Documents
Machines vs humans
Among the five senses, vision is
considered to be vital one for a human
being.
But a human being can perceive only
visible part of the electromagnetic
spectrum.
But machines can span the entire range of
electromagnetic spectrum from gamma to
radio waves.
Categorizing Images
One can categorize images according to their source.
One of the main sources is due to electromagnetic
spectrum.
X-ray Imaging
It is also used in medicine and astronomy.
We can get images of blood vessels in
angiography.
It is also used in Computerized Axial
Tomography (CAT) to generate 3-D
rendition of a patient.
High energy X-ray images are used in
industrial processes (electronic circuit
board).
Ultraviolet Imaging
It is used in lithography, industrial
inspection, microscopy, lasers, biological
imaging and astronomical observations.
Corn, cereals, onions caused by parasitic
fungi can be identified using this imaging
technique.
Microwave Images
(Thumb print and paper currency)
Radar images belong to this category.
It can penetrate the inaccessible
regions of earths surface.
Radio Images
(MRI of human knee and spine)
It is used in Magnetic Resonance Imaging (MRI).
Acoustic imaging
The images use sound energy.
They are used in geological exploration,
industry and medicine.
It is also used in mineral and oil
exploration.
Ultrasonic images are used in obstetrics to
determine the health of unborn babies and
determining the sex of the baby.
Components of an Image
Processing System
Specialized Image Processing Hardware It consists of digitizer plus hardware that performs
other primitive operations such as an arithmetic logic
unit (ALU), which performs arithmetic and logical
operations on entire image.
This type of hardware is also called as front-end
subsystem and its characteristic is speed.
This unit does things that require fast data throughputs
which main computer cannot handle.
Computer In an image processing system it is a
general-purpose computer.
Software It consists of specialized modules that does
specific tasks (eg. matlab)
Brightness Adaptation
There are 2 phenomena which clearly
demonstrate that the perceived brightness
is not a simple function of intensity.
The visual system tends to undershoot or
overshoot around the boundary of regions
of different intensities.
These scalloped bands near the
boundaries are called Mach bands.
0 f ( x, y )
0 r ( x, y ) 1
Image Sampling
The output of many sensors is continuous
voltage.
To get a digital image, we need to convert this
voltage into digital form.
But this involves 2 processes, namely sampling
and quantization.
An image is continuous with respect to x and y
coordinates and in amplitude.
Digitizing the coordinates is called sampling.
Digitizing the amplitude is called quantization.
250
200
150
100
50
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50
60
70
80
90
100
f(0,0) f(0,1)
f(0,N-1)
f(1,0) f(1,1)
f(1,N-1)
f(x,y) =
.
.
.
:
:
:
b M N k
When M = N, b = N^k
For example, a 256 X 256 image represented
with 8 bits takes 5,24,288 bits.
Subsampling
The subsampling is achieved by deleting
appropriate number of rows and columns from
the original image.
For example, we can get a 256 X 256 image by
deleting the every other row and column from
the 512 X 512 image.
In order to see the effect of subsampling, we can
replicate appropriate rows and columns of 256 X
256 to bring it to 512 X 512 size.
We notice the checkboard pattern.
Bilinear interpolation
The best way is to use bilinear interpolation
using four nearest neighbors of a point.
Let (x,y) denote the coordinates of a point in the
zoomed image and let v(x,y) denote the gray
level assigned to it.
The assigned gray level is given by
v(x,y) = ax + by + cxy + d
Here the four coefficients are determined from
the 4 equations in four unknowns that can be
written using the 4 nearest neighbors of point
(x,y). (Assignment)
Shrinking an image
For shrinking an image by one-half, we delete
every other row and column.
In order to shrink an image by non integer factor,
we expand the grid to fit over the original image,
do gray-level nearest neighbor or bilinear
interpolation, and then shrink the grid back to its
original specified size. (Assignment).
It is good to blur an image slightly before
shrinking it.
Image size:256X256
Image size:512X512
Image size:256X256
Image size:512X512
3 types of adjacency
q is in N4(p), or
Digital Path
Arrangement of pixels
Distance Measures
For pixels p, q and z with coordinates (x,y), (s,t) and
(v,w) respectively, D is a distance function or metric if
D(p,q) 0
(D(p,q) = 0 iff p = q)
D(p,q) = D(q,p) and
D(p,z) D(p,q) + D(q,z)
The Euclidean distance between p and q is defined as
D4 distance
The D4 distance (called city-block distance) between p
and q is defined as
D4(p,q) = | x s| + | y t|.
Here the pixels have a D4 distance from (x,y) less than or
equal to some value r form a diamond centered at (x,y).
(eg) the pixels with D4 distance 2 from (x,y) (the center
point) form the following contours of constant distance:
2
1
2
2
1
0
1
2
2
1
2
2
The pixels with D4 = 1 are the 4-neighbors of (x,y).
D8 distance
The D8 distance (called chessboard distance) between p
and q is defined as
2
2
2
2
2
2
1
1
1
2
2
1
0
1
2
2
1
1
1
2
2
2
2
2
2
The pixels with D8 = 1 are the 8-neighbors of (x,y).