COM2304: Intensity Transformation and Spatial Filtering – II Spatial Filtering and Smoothing

COMPUTER GRAPHICS & IMAGE PROCESSING
COM2304
Intensity Transformation and Spatial Filtering – II
Spatial Filtering and Smoothing
K.A.S.H.Kulathilake
B.Sc. (Hons) IT (SLIIT), MCS (UCSC), M.Phil (UOM), SEDA(UK)
Rajarata University of Sri Lanka
Faculty of Applied Sciences
Department of Physical Sciences

Learning Outcomes
COM2304 - Computer Graphics & Image
Processing
• At the end of this lecture, you should be
able to;
– describe the fundamentals of spatial filtering.
– generating spatial filter masks.
– identify smoothing via linear filters and non
linear filters.
– apply smoothing techniques for problem
solving.

Filters in Image Processing
• Filter can accept or reject certain frequency
components.
• In image processing filters are mainly used to
suppress either the high frequencies in the
image, i.e. smoothing the image, or the low
frequencies, i.e. enhancing/ sharpening or
detecting edges in the image.
• An image can be filtered either in the
frequency or in the spatial domain.
Processing

Fundamentals of Frequency
Domain Filtering
• Frequency domain filtering involves ;
– transforming the image into the
frequency domain,
– multiplying it with the frequency
filter function and
– re-transforming the result into the
spatial domain.
• The filter function is shaped so as to
attenuate some frequencies and
enhance others.
• For example, a simple lowpass function
is 1 for frequencies smaller than the
cut-off frequency and 0 for all others.
Processing

Fundamentals of Spatial Filtering
• Spatial filter contains two components;
– MASK/ CONVOLUTION MATRIX/ KERNEL: A neighborhood
(typically small rectangle:)
– FILTER FUNCTION: A predefined operation that is performed
on the image pixels encompassed by the neighborhood.
• Spatial filtering is the method of choice in situations when only
additive random noise is present.
• Filtering creates a new pixel value with coordinates equal to the
coordinates of the center of the neighborhood, and whose value
is the result of the filtering operation.
Processing

(Cont….)
Processing

(Cont….)
Processing
If the operation
performed on the image
pixels is linear, then the
filter is called a linear
spatial filter, otherwise,
the filter is non-linear.

(Cont….)
• Spatial filtering of an image of size M×N with a filter of size
m×n is given by the expression;
• Where w is the mask and f is the image function, x and y
are varied so that each pixel in w visits every pixel in f.
• m = 2a+1 and n = 2b+1, where a and b are positive
integers.
• This concept is known as correlation.
Processing

(Cont….)
• There are two closely related concepts that
must be understood clearly when performing
linear spatial filtering
– Correlation
• Correlation is the process of moving a filter mask over
the image and computing the sum of products at each
location exactly as explained previously.
– Convolution
• The mechanics of convolution are the same as
correlation except that the filter is first rotated by 180
degrees.
Processing

(Cont….)
Processing
If the filter mask is symmetric, correlation and convolution
yield the same result

Mask/Kernel/ Convolution Matrix
• In image processing, a kernel, convolution
matrix, or mask is a small matrix.
• It is useful for blurring, sharpening,
embossing, edge detection, and more.
• This is accomplished by means of convolution
between a kernel and an image.
Processing

(Cont…)
• Important Features of
Kernel
– Size: 3×3, 5×5, 9×9,
21×21
– Shape: rectangle, cross,
strip, circular, diamond
or user defined.
– Coefficients / Values:
set based on the
operation.
– Anchor point: mostly in
middle
Processing
0 -1 0
-1 5 -1
0 -1 0
3×3 Kernel
Anchor
Values for
sharpening
operation

(Cont…)
Processing
0 -1 0
-1 5 -1
0 -1 0
Kernel

(Cont…)
Processing
http://setosa.io/ev/image-kernels/
https://en.wikipedia.org/wiki/Kernel_(image_processing)

Convolution
• Convolution is a simple mathematical
operation which is fundamental to many
common image processing operators.
• This can be used in image processing to
implement operators whose output pixel
values are simple linear combinations of
certain input pixel values.
Processing

Convolution (Cont…)
Processing
Each kernel position
corresponds to a single
output pixel, the value of
which is calculated by
multiplying together the
kernel value and the
underlying image pixel
value for each of the cells
in the kernel, and then
adding all these numbers
together.
The convolution is
performed by sliding the
kernel over the image,
generally starting at the
top left corner.

Convolution (Cont…)
Processing

Generating Spatial Filter Masks
• Generating an m×n linear spatial filter requires that we
specify mn mask coefficients.
• These coefficients are selected based on what the filter
is supposed to do.
• There are different types of filters available, some of
them are;
– Linear filters - In signal processing, a linear filter is a filter
whose output is a linear function of its input.
– Continuous function based filters – e.g. Gaussian
smoothing (it’s a kind of linear filter).
– Non linear filters - In signal processing, a non-linear filter is
a filter whose output is not a linear function of its input.
Processing

Generating Spatial Filter Masks (cont…)
• Linear filters
– Implement sum of products
– ex: filters used for smoothing
• Continuous function based filters
– ex: Gaussian function of two variables as shown in the following
equation;
– Where σ is the standard deviation and coordinates x and y are
the positive integers.
– Recall that 2D Gaussian function has bell shape and the σ
control the tightness of the bell.
Processing

Generating Spatial Filter Masks (cont…)
• Non linear/ Order static filters
– Non linear filters require to specify the size of the
filter and the operation(s) to be performed on the
image pixel contained in the filter.
• ex: filter 5×5 filter centered at an arbitrary point and
perform max operation.
– Non linear filters are quiet powerful in some
applications, than the linear filters.
Processing

Smoothing Spatial Filters
• Smoothing filters are used for blurring and
noise reduction.
• Blurring is used in preprocessing tasks, such as
removal of small details from an image prior
to object extraction, and bridging of small
gaps in lines or curves.
• Noise reduction can be accomplished by
blurring with a linear filter and also by
nonlinear filtering.
Processing

Smoothing Linear Filters
• Average /Mean Filter
– The output is simply the average of the pixels
contained in the neighborhood of the filter mask.
– In smoothing filter, it replaces the value of every pixel
in an image by the average of the intensity levels in
the neighborhood defined by the filter mask.
– This process result in an image with reduced “sharp”
transitions in intensities.
– Most obvious application of smoothing is noise
reduction because random noise typically consists of
sharp transitions in intensity levels.
Processing

Smoothing Linear Filters (Cont…)
Processing

– Averaging filters blur edges, because edges
represent the sharp intensity transition.
– Average blurring smoothes the false contours that
result from using an insufficient number of
intensity levels.
– A major use of averaging filters is in the reduction
of “irrelevant” detail (pixel regions that are small
with respect to the size of the filter mask) in an
image.
Processing

Processing
Observe
blurring
occur during
smoothing
operation

– Let Sx,y represent the set of coordinates in neighbourhood
of size m×n, centered at point (x, y).
– The arithmetic mean filter computes the average value of
the corrupted image g( x, y) in the area defined by Sx,y.
– The value of the restored image f’ at point ( x, y) is simply
the arithmetic mean computed using the pixels in the
region defined by Sx,y.
Processing


xySts
tsg
mn
yxf
),(
),(
1
),(ˆ

• There are different kinds of mean filters all of
which exhibit slightly different behaviour:
• Geometric Mean:
– Achieves similar smoothing to the arithmetic
mean, but tends to lose less image detail
Processing
mn
Sts xy
tsgyxf
1
),(
),(),(ˆ








 

• Harmonic Mean:
– Works well for salt noise, but fails for pepper
noise.
– Also does well for other kinds of noise such as
Gaussian noise
Processing


xySts tsg
mn
yxf
),( ),(
1
),(ˆ

• Contraharmonic Mean:
– Q is the order of the filter and adjusting its value
changes the filter’s behaviour.
– Positive values of Q eliminate pepper noise.
– Negative values of Q eliminate salt noise.
– If Q =0 it represents the arithmetic filter.
Processing






xy
xy
Sts
Q
Sts
Q
tsg
tsg
yxf
),(
),(
1
),(
),(
),(ˆ

Original
Image
Image
Corrupted
By Gaussian
Noise
After A 3*3
Geometric
Mean Filter
After A 3*3
Arithmetic
Mean Filter
Processing

Image
Corrupted
By Pepper
Noise
Result of
Filtering Above
With 3*3
Contraharmonic
Q=1.5
Processing

Image
Corrupted
By Salt
Noise
Result of
Filtering Above
With 3*3
Contraharmonic
Q=-1.5
Processing

Choosing the wrong value for Q when using the
contraharmonic filter can have drastic results
Processing

• A spatial averaging filter in which all
coefficients are equal is called a Box
Filter.
• The basic strategy behind weighting
the center point the highest and then
reducing the value of the coefficients
as a function of increasing distance
from the origin in spatial averaging
filter is simply an attempt to reduce
blurring in the smoothing process.
Processing

• Gaussian Smoothing:
– The Gaussian filter is a non-uniform low pass filter.
– The kernel coefficients diminish with increasing distance from
the kernel’s centre.
– Central pixels have a higher weighting than those on the
periphery.
– Larger values of σ produce a wider peak (greater blurring).
– Kernel size must increase with increasing σ to maintain the
Gaussian nature of the filter (`bell-shaped') .
– Gaussian kernel coefficients depend on the value of σ.
– At the edge of the mask, coefficients must be close to 0.
– The kernel is rotationally symmetric with no directional bias.
– Gaussian filters might not preserve image brightness.
Processing

• The Gaussian distribution
in 1-D has the form:
• Where σ is the standard
deviation of the
distribution.
• We have also assumed
that the distribution has a
mean of zero (i.e. it is
centered on the line x=0).
Processing
1-D Gaussian distribution with
mean 0 and σ=1

• In 2-D, an isotropic
(i.e. circularly symmetric)
Gaussian has the form:
Processing
2-D Gaussian distribution with
mean (0,0) and σ=1
Discrete approximation to Gaussian
function with σ=1
The std. dev σ of the Gaussian
determines the amount of
smoothing.

Smoothing Order Static (Nonlinear)
Filters
• The response of the order static or nonlinear
filters is based on ordering (ranking) the pixels
contained in the image area encompassed by the
filter, and then replacing the value of the center
pixel with the value determine by the ranking
result.
• Useful nonlinear filters include;
– Median filter
– Max and min filter
– Midpoint filter
Processing

Filters (Cont…)
• Median Filter:
– It can reduce certain typed of random noise with
less blurring than the linear smoothing filters of
similar size.
– Provides excellent results when applying to reduce
salt and pepper noise.
Processing
)},({),(ˆ
),(
tsgmedianyxf
xySts 


Filters (Cont…)
Processing

Filters (Cont…)
• Median filter smoothes ("washes") all edges
and boundaries and may "erase" all details
whose size is about n/2× m/2, where n × m is
a window size.
• As a result, an image becomes "fuzzy".
• Median filter is not so efficient for additive
Gaussian noise removal, it yields to linear
filters.
Processing

Filters (Cont…)
• Max Filter:
• Min Filter:
–Max filter is useful for finding the brightest points in an
image and min filter for finding the darkest points in an
image.
–Max filter is good for pepper noise and min filter is good for
salt noise
Processing
)},({max),(ˆ
),(
tsgyxf
xySts 

)},({min),(ˆ
),(
tsgyxf
xySts 


Filters (Cont…)
• Midpoint Filter:
– Good for random Gaussian and uniform noise
Processing



 

)},({min)},({max
2
1
),(ˆ
),(),(
tsgtsgyxf
xyxy StsSts

Filters (Cont…)
• Alpha-trimmed Mean Filter:
– We can delete the d/2 lowest and d/2 highest grey
levels.
– So g(s, t) represents the remaining mn – d pixels.
– When d= 0 : arithmetic mean filter.
– Good for combination of salt and pepper and
Gaussian noise.
Processing


xySts
tsg
dmn
yxf
),(
),(
1
),(ˆ

Removing Periodic Noise
• Removing periodic noise form an image
involves removing a particular range of
frequencies from that image.
• Band reject filters can be used for this
purpose.
Out of the scope of this course -   
Processing

Reference
• Chapter 03, Chapter 05 of Gonzalez, R.C.,
Woods, R.E., Digital Image Processing, 3rd ed.
Addison-Wesley Pub.
• http://setosa.io/ev/image-kernels/
• https://en.wikipedia.org/wiki/Kernel_(image_
processing)
Processing

Learning Outcomes Revisit
• Now, you should be able to;
– describe the fundamentals of spatial filtering.
– generating spatial filter masks.
– identify smoothing via linear filters and non
linear filters.
– apply smoothing techniques for problem
solving.
Processing

QUESTIONS ?
Next Lecture – Intensity Transformation and Spatial Filtering – III
Processing

COM2304: Intensity Transformation and Spatial Filtering – II Spatial Filtering and Smoothing

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to COM2304: Intensity Transformation and Spatial Filtering – II Spatial Filtering and Smoothing (20)

More from Hemantha Kulathilake (20)

Recently uploaded (20)

COM2304: Intensity Transformation and Spatial Filtering – II Spatial Filtering and Smoothing