0% found this document useful (0 votes)
13 views59 pages

4 - Unit DIP

Uploaded by

mohanmysore6742
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
13 views59 pages

4 - Unit DIP

Uploaded by

mohanmysore6742
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 59

Unit - 4

Image Restoration
Syllabus
A model of Image degradation and Restoration
• Restoration : is the process of inverting a degradation using
knowledge about the nature.
• Whereas enhancement is a process that aims to improve ‘bad’ quality
image so that it looks better.
• Degradation : Becomes poor quality due to noise, blurring and
Artifacts.

• Restoration: Bring it back to a good quality. Restoring image as if


there is no degradation in the image.

• Image recovering : Recovering the image which is degraded by some


knowledge of degradation function H.
What is a noise
• Noise tells unwanted information in digital images. Noise produces
undesirable effects such as artifacts, unrealistic edges, unseen lines,
corners, blurred objects and disturbs background scenes. To reduce
these undesirable effects, prior learning of noise models is essential
for further processing.
Why noise model?
• The benefit of understanding the noise model is that we can easily
identify the noise type and apply filters to remove the noise.
Noise and Image
• The source of noise in digital images arise during image acquisition
and transmission.
• Image sensor can be affected by ambient conditions
• Interference can be added to an image during transmission
• To understand the restoration process it is important to understand
the noise function n(x,y)
• Noise cannot be predicted but can be approximately described in
statistical way using the probability density function PDF
Noise model
• The different noise model are:
• Gaussian Noise Model
• Rayleigh Noise
• Erlang Model
• Exponential Noise
• Uniform Noise
• Salt and Pepper noise (Impulse)
Gaussian Noise
• It is the most popular noise model
• It is having a bell shaped function
• We need to compute mean, variance and standard deviation.
• It enters the system and effects both dark and light region.
• The Gaussian model is shown in next slide
Gaussian Noise or Normal Noise model
Reyleigh noise Model
• Little similar to Gaussian model, but it is not simitric on both side
• Left portion is abrupt.
• It is skewed to the right side
• It is usual in remote sensing applications
Rayleigh Noise Model
Erlang or Gamma Noise
• It is little similar to Rayleigh Noise
• But the left side starts from zero.
• It occurs due to illumination problem.
Erlang or Gamma Noise
Exponential Noise
• It is exponential decaying
• Usually seen in laser imaging
Exponential Noise
Uniform Noise
• It is uniform noise over certain range.
• It occurs due to quantization problem
Uniform Noise
Salt and Pepper noise
• It is also called impulse error.
• It occurs due to sensor and memory error
Salt and Pepper Noise
Example: for Noise model : Figure shows the
levels of colour distribution
Gaussian, Reyleigh and Erlang Noise
Exponential, Uniform and Impulse Noise
Advantage of noise model
• If we take the image and histogram of the image, we can judge what
type of noise is added to the image.
• Based on this suitable filter can be designed.
Image Restoration in presence of noise only -
special filtering
• If the degradation is due to blurring or artifact then frequency
filtering is used
• If only noise is present, then spatial filtering is the method of
removing the noise.
• Noise is unwanted information.
• In spatial filtering, we apply mask on the image.
Model is to be changed, as only noise is
present. Here type of noise is also not known.
Mean (addition) and Geometric (Multiply)
filter
Harmonic and Contra harmonic Filter
Contraharmonic filter : Example
Order Statistics Filter
Different order statistics filters
Alpha Trimmed Mean Filter
Periodic Noise Reduction using Frequency Domain
Band Reject Filter
• A band reject filter is useful when the general location of the noise in
the frequency domain is known. A band reject filter blocks
frequencies within the chosen range and lets frequencies outside of
the range pass through.
Ideal Band Reject Filter
Butterworth Band Reject
Gaussian Band Reject
Example: Band Reject
Quick recap of different high pass filter…
Summary of Band Reject Filter
Band Pass filter
Notch Filter
Notch –Ideal filter
Notch Filter
Linear, Position invariant Degradations
Estimating of Degradation Function
• Degradation function can be estimated in three methods
• Estimation by Observation
• Estimation by Experimentation
• Estimation by Modelling
Estimation by observation
• Degradation function (H) can be estimated by visually looking into
section of the image
• Details in next slide
Estimating Degradation Function
Inverse Filtering
What is wiener filter?
• The Wiener filter is the MSE-optimal stationary linear filter for images
degraded by additive noise and blurring.
Wiener Filter
End of Unit 4

You might also like