Image Processing Image Processing: Intensity Transformations Intensity Transformations CH 3 CH 3 Chapter 3 Chapter 3

Image
Processing
Intensity Transformations
Ch
Chapter 33
Prof. Vidya Manian
Dept. of Electrical and Comptuer Engineering
INEL 5327 ECE, UPRM Intensity Transformations ‐ 1

Overview
• Background
• Basic intensity transformation functions
• Histogram processing
i i
INEL 5327 ECE, UPRM 2
3x3 neighborhood of a point (x,y)
3x3 neighborhood of a point (x,y)
Intensity or gray‐level or mapping
transformation function
f f
• S=T(r)
• Resulting image has higher contrast by
darkening the intensity levels below k and
darkening the intensity levels below k and
brightening the levels above k
• Contrast stretching‐values of r lower than k
C hi l f l h k
are compressed by the transformation
f
function into a narrow range of s, toward
i i f d
black, vice versa for values of r higher than k
Contrast stretching and thresholding
functions
Thresholding function
• T(r)
T(r) produces a two
produces a two‐level
level (binary) image
(binary) image –
limiting case
• Intensity transformations are used for image
Intensity transformations are used for image
enhancement
• Also for image segmentation
Al f i i
• Two types: point processing and
neighborhood processing
Basic intensity transformation
functions
• S=T(r)
• Mapping from r to s are implemented via a
lookup table
lookup table
• For 8‐bits, lookup table has 256 entries.
• Types of functions: linear (negative and
identity transformations)
• Power‐law (nth power and nth root
transformations))
Basic intensity transformation
functions
Image negatives
Image negatives
• Negative transformation s=L‐1‐r
g
• Log transformation: s=c log (1+r)
• c is a constant, r≥0
• Maps range of low intensity values in the input to
wider range of output levels
• Expands dark pixels in an image and compresses
E d d k i l i i d
the higher‐level values (white pixels)
• Useful for Fourier spectrum: spectra values range
Useful for Fourier spectrum: spectra values range
from 0 to 10^6 or higher.
• Detail lost in the display of Fourier spectrum
Original image and negative image
obtained using the negative
obtained using the negative
transormation
Fourier spectrum and log
transformation
f
• Fourier
Fourier spectrum with values in range of 0 to
spectrum with values in range of 0 to
1.5 x 10^6 (Fig. 3.5 a) only few pixels are
white.
white
• First use log transformation with c=1, range of
values become 0 to 6 2
values become 0 to 6.2
• Fig 3.5b has better scaling and shows more
d il
detail.
Power law (Gamma) transformations
Power law (Gamma) transformations
• s=cr
Where c and  are positive constants.
• s=c(r+) to account for an offset due to display
calibration
• power‐law curves with fractional values of  map a
narrow range of dark input values to a wider range of
output values, and opposite for higher values of input
t t l d it f hi h l fi t
levels
• >1 has opposite effect of <1. Also called Gamma
 pp 
correction
• Device dependent, CRT displays, image capture,
printing
• Preprocess image before input to monitor by
p g p y
s=r^(1/2.5)=r^(0.4)
• Gamma corrected output has appearance close
to original
to original
• Applicable to scanners, printers
• Use of digital images for commercial purpose
Use of digital images for commercial purpose
over Internet has increased
• Solution –display image after gamma correction
p y g g
to value that represents “average” of the types of
monitors and computer systems to be used to
display the image
display the image.
Contrast stretching
Contrast stretching
Bit plane representation of an 8‐bit
image
Four basic image types
Histogram equalization
Histogram equalization
Transformation functions for
Histogram equalization in Fig. 3. 20
l
Example of histogram equalization
Example of histogram equalization
Histogram
Histogram equalized image
Histogram equalized image
Specified Histogram
Specified Histogram
Steps for Histogram Matching
Steps for Histogram Matching
• Obtain Pr(r) from input image and obtain values of s
• Use specified pdf
Use specified pdf to obtain the transformation function
to obtain the transformation function
G(z)
• Obtain the inverse transformation z=G‐1(s); mapping
( ); pp g
from s to z
• Obtain output image by first equalizing input image .
F
For each pixel with value s, perform inverse mapping
h i l ith l f i i
z=G‐1(s) to obtain corresponding output pixel
• When all pixels are processed the pdf
When all pixels are processed the pdf of output image
of output image
is equal to specified pdf
Local Histogram equalization
Local Histogram equalization
Global and local histogram
equalization
l
Fundamentals of spatial filtering
Fundamentals of spatial filtering
• Filter:
Filter: accepting(passing) or rejecting certain
accepting(passing) or rejecting certain
frequency components
• Spatial filter consists of (1) a neighborhood (a
Spatial filter consists of (1) a neighborhood (a
small rectangle) and (2) predefined operation
performed on the image pixels encompassed
performed on the image pixels encompassed
by the neighborhood
• Linear and non‐linear filters
Li d li fil
Basics of Spatial Filtering
Basics of Spatial Filtering
• Neighborhood
Neighborhood subimage
subimage‐filter,
filter, mask, kernel,
mask, kernel,
template, or window
• Values in mask
Values in mask –coefficients
coefficients
• Response R of linear filtering with the filter
mask at a point (x,y) in the image is:
mask at a point (x,y) in the image is:
• R=w(‐1,‐1)f(x‐1,y‐1)+w(‐1,0)f(x‐1,y)+…
• +w(0,0)f(x,y)+…+w(1,0)f(x+1,y)+w(1,1)f(x+1,y+
+w(0 0)f(x y)+ +w(1 0)f(x+1 y)+w(1 1)f(x+1 y+
1)
Mechanics of spatial filtering
Mechanics of spatial filtering
• W(0
W(0,0) center coefficient of the filter
0) center coefficient of the filter
• Mask of size mxm, m=2a+1 and n=2b+1,
where a and b are positive integers
where a and b are positive integers
• Smallest size 3x3
Spatial correlation and convolution
Spatial correlation and convolution
• Co
Correlation is the process of moving a filter mask
e at o s t e p ocess o o g a te as
over the image and computing the sum of
products at each location.
• Convolution is the same except that the filter is
rotated by 180 degrees.
• Rotation is equivalent to horizontal flipping
• In 2Drotation is equivalent to flipping the mask
along one axis and then the other
l i d th th th
• Fig. 3.29 shows 1‐D convolution and correlation
Correlation and convolution
Correlation and convolution
Smoothing spatial filters
Smoothing linear filters
Smoothing linear filters
• An
An mxn
mxn mask would have a normalizing
mask would have a normalizing
constant equal to 1/mn.
• A spatial averaging filter in which all
A spatial averaging filter in which all
coefficients are equal is called a box filter
• The diagonal terms are further away from
The diagonal terms are further away from
center and hence are weighted less than the
immediate neighbors
• Sum of coefficients =16 is useful for easy
computer implementation ( a power of 2)
Effect of smoothing as a function of
fl
filter size
• Average
Average filters of sizes m
filters of sizes m=3,5,9,15
3,5,9,15 and 35
and 35
pixels
• Larger size windows for blurring removes
Larger size windows for blurring removes
smaller objects from image
• A gross representation of objects of interest is
A gross representation of objects of interest is
obtained
• Intensity of smaller objects blends with
Intensity of smaller objects blends with
background and larger objects become
bloblike
Order‐statistics
Order statistics (nonlinear) filters
(nonlinear) filters
• Response of nonlinear filters based on
p
ordering(ranking) the pixels contained in the
image area encompassed by the filter
• Median filter‐
Median filter to remove impulse noise (salt and
to remove impulse noise (salt and
pepper noise)
• Sort the pixel values in the window (say 3x3 ) and
Sort the pixel values in the window (say 3x3 ) and
assign the median value to the center pixel
• Forces pixels with distinct intensity levels to be
more like its neighbors
lik i i hb
• Max, min filters
Image filtered with a 15x15 averaging
maskk
Sharpening spatial filters
• Highlights transition in intensities
Highlights transition in intensities
• Blurring by pixel averaging
• Sh
Sharpening by spatial differentiation
i b i l diff i i
• Enhances edges and other discontinuities such
as noise
• Based on first and second order derivatives
First and second derivatives of a 1D
signall
f
 f ( x  1)  f ( x )
x
2 f
 f ( x  1)  f ( x  1)  2 f ( x )
x
x 2
Change between adjacent pixels
First and second derivative
First and second derivative
(1) Must be zero in flat areas;
(2) Non zero at the onset of gray‐level step or ramp
(3) First derivative‐non zero along ramps; second derivative‐
Zero along ramps of cnstant slope
The Laplacian for enhancement (the
second derivative)
dd i i )
2 f 2 f
 f  2  2
2
x y
In the x direction
2 f
 f ( x  1, y )  f ( x  1, y )  2 f ( x, y )
x 2
In the y direction
2 f
 f ( x, y  1)  f ( x, y  1)  2 f ( x, y )
x 2
Sum min g
2 f  [ f ( x, y  1)  f ( x, y  1)  f ( x  1, y )  f ( x  1, y )]  4 f ( x, y )
Second derivative‐the
Second derivative the Laplacian
Laplacian
Blurring using Gaussian filter
Blurring using Gaussian filter
The Gradient for enhancement (the first
d i i )
derivative)
 f 
Gx   x 
f      
G y   f 
 y 
f  mag (f )
 [Gx 2  G y 2 ]1/ 2
2 1/ 2
 f   f  
2
      
 x   y  
f  ( z7  2 z8  z9 )  ( z1  2 z1  z3 )  ( z3  2 z6  z9 )  ( z1  2 z4  z7 )
Composite Laplacian mask
p p
Sobel gradient
Exercise in group
Exercise in group
• Propose
Propose a set of intensity
a set of intensity‐slicing
slicing
transformations capable of producing all the
individual bit planes of an 8bit monochrome
image. T(r)=0 in range [0,127] and T(r)=255 for
r in range [128,255] produces an image of the
8th bit plane in an 8‐bit image.
bi l i 8 bi i
• Given the Gaussian pdf what is the
transformation function you would use for
f i f i ld f
histogram equalization.
Exercises on spatial filtering
Exercises on spatial filtering
• 3.1,3.2,3.4, 3.5, 3.6, 3.7, 3.8 to 3.11, 3.13
3 1 3 2 3 4 3 5 3 6 3 7 3 8 to 3 11 3 13
• 3.15, 3.17, 3.19,
• 3.21, 3.23, 3.24, 3.25, 3.27, 3.28, 3.29
• 3
3.1)Give a single intensity transformation
1)Give a single intensity transformation
function for spreading the intensities of an
image so the lowest intensity is 0 and highest
image so the lowest intensity is 0 and highest
is L‐1.
• 3.5) What effect would setting to zero the
3 5) What effect would setting to zero the
lower‐order bit planes have on the histogram
of an image in general
of an image in general.
• 3
3.6) Explain why discrete histogram
6) Explain why discrete histogram
equalization does not in general yield a flat
histogram.
histogram
• Answer: To get a flat histogram, pixel
intensities should be redistributed so that
intensities should be redistributed so that
there are L groups of n/L pixels with same
intensity and n=MN
intensity and n=MN.
• 3
3.7) Suppose that a digital image is subjected
7) Suppose that a digital image is subjected
to histogram equalization. Show that a second
pass of histogram equalization (on the
pass of histogram equalization (on the
histogram equalized image) will produce
exactly the same result as the first pass
exactly the same result as the first pass.

Image Processing Image Processing: Intensity Transformations Intensity Transformations CH 3 CH 3 Chapter 3 Chapter 3

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Image Processing Image Processing: Intensity Transformations Intensity Transformations CH 3 CH 3 Chapter 3 Chapter 3

Uploaded by

Copyright:

Available Formats

Image

INEL 5327 ECE, UPRM Intensity Transformations ‐ 1

You might also like