0% found this document useful (0 votes)
16 views

Module 4

The document provides an overview of image restoration techniques. It discusses that image restoration attempts to recover an image that has been degraded using prior knowledge of degradation phenomena. It is an objective process unlike enhancement which is subjective. Common sources of noise include sensor noise during image acquisition and interference during transmission. Key techniques discussed include inverse filtering, Wiener filtering, and non-linear methods like trimmed average filtering and median filtering. Applications mentioned are in medical imaging, astronomy, printing, and defense. Noise models covered are Gaussian, uniform, additive, and salt-pepper noise.

Uploaded by

suma_hari6244
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
16 views

Module 4

The document provides an overview of image restoration techniques. It discusses that image restoration attempts to recover an image that has been degraded using prior knowledge of degradation phenomena. It is an objective process unlike enhancement which is subjective. Common sources of noise include sensor noise during image acquisition and interference during transmission. Key techniques discussed include inverse filtering, Wiener filtering, and non-linear methods like trimmed average filtering and median filtering. Applications mentioned are in medical imaging, astronomy, printing, and defense. Noise models covered are Gaussian, uniform, additive, and salt-pepper noise.

Uploaded by

suma_hari6244
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 36

ECT 385: Topics in digital image

processing
S5 Minor
Module 4 : Image restoration

Reference Textbook: Digital Image Processing by S Jayaraman, S Essakirajan, T Veerakumar,


Digital Image Processing by Rafael C. Gonzalez and Richard E. Woods
Image restoration
• Image restoration attempts to recover an image that has been degraded using a prior knowledge of degradation
phenomenon.
• It is an objective process unlike enhancement which is subjective.
• The principal sources of noise in digital images arise during image acquisition and/or transmission.
• The performance of imaging sensors is affected by a variety of environmental factors during image acquisition, and
by the quality of the sensing elements themselves.
• For instance, in acquiring images with a CCD camera, light levels and sensor temperature are major factors
affecting the amount of noise in the resulting
image.
• Images are corrupted during transmission principally by interference in the transmission channel.
• For example, an image transmitted using a wireless network might be corrupted by lightning or other
atmospheric disturbance.
• Types of degradation: relative camera-object motion, blur due to camera misfocus, atmospheric turbulence
• Some restoration techniques are formulated in spatial domain (e.g. additive noise) while others are in frequency domain
(e.g. using filter mask)
Image restoration Vs Image enhancement
Model of Image Degradation/Restoration

• Restoration produces an estimate ḟ (𝑥, 𝑦 ) of the original image from the


degraded image 𝑔 ( 𝑥 , 𝑦 )

• In spatial domain, the degraded image is given by:


• The estimate ḟ (𝑥, 𝑦) of the image after restoration is given by

• In frequency domain, the degraded image is given by:

• The estimate of image after restoration


Image restoration techniques
• Image-restoration technique can be broadly classified into two types depending
upon the knowledge of degradation.
• If the prior knowledge about degradation is known then the deterministic method
• If it is not known then the stochastic method of image restoration
Types of Linear RestorationModels

Unconstrained Restoration:

No a priori knowledge about the noise is assumed.

E.g. Inverse Filter, Pseudo Inverse Filter

Constrained Restoration:

Some a priori knowledge of the noise is used.

E.g. Weiner Filter, Constrained Least Square Filtering

Rajagiri School of Engineering & Technology, Kochi 7


Pseudo
inverse filter

Weiner
filter
Pseudo-inverse filter
• Mostly, H(u, v) represents is a low-pass filter, which implies that H(u, v) ≈ 0 at high frequencies.

• The division of H(u,v) leads to large amplification at high frequencies, where the noise dominates over the
image.

• This frequency-dependent amplification leads to significant errors in the restored image, and amplification of
noise.

• Hence a pseudo-inverse filter is defined as

• The value of ε affects the restored image.


Non-linear restoration methods
Trimmed Average filter
• The average filter at a spatial location (x, y) on the image is defined by finding the average of the interested
pixel (centre pixel) with its 8 (if 3 × 3 window is used) near neighbours.
• The centre pixel is assigned to the average value. The same averaging is applied all over the image for all pixels which
results in the averaging filter operation of the image.
• The significance of average filtering lies in its impulse-noise-suppression capability.
• Along with noise removal, it creates a blurring effect as an unintended side effect.
• Blurring can be controlled by reducing the number of ‘interested’ neighbour pixels.
• The solution for blurring converges towards ‘eliminating’ a few neighbouring pixels based on the statistical relationships
for the calculation of the average value.
• Objectives converge towards trimming the ‘un-interested pixels’ from the centre pixel based on the statistical criteria.
• Trimming Criteria
• The extreme minimum value and extreme maximum value in the windowed pixels will affect the average value
calculated. So by gradually trimming the pixels from both minimum extreme and maximum extreme, the optimum
performance can be achieved. Consecutive extreme values were identified and trimmed by ordering the windowed
pixels in ascending order.
Non-linear restoration methods
• Median filter
• The best-known order-statistic filter in image processing is the median filter, which, as its name implies,
replaces the value of a pixel by the median of the intensity levels in a predefined neighborhood of that
pixel:

• where, as before, S xy is a subimage (neighborhood) centered on point (x, y).

• The value of the pixel at (x, y) is included in the computation of the median.

• Median filters are quite popular because, for certain types of random noise, they provide excellent noise-
reduction capabilities, with considerably less blurring than linear smoothing filters of similar size. Median
filters are particularly effective in the presence of both bipolar and unipolar impulse noise
• In order to perform median filtering at a point in an image,
• we first sort the values of the pixels in the neighborhood,
• determine their median,
• assign that value to the pixel in the filtered image corresponding to the center of the neighborhood.
• For example, in a 3 x 3 neighborhood the median is the 5th largest value, in a 5 x 5 neighborhood it is the
13th largest value, and so on.
• The principal function of median filters is to force points to be more like their neighbors.
• Isolated clusters of pixels that are light or dark with respect to their neighbors, and whose area one-half
the filter area, are forced by an m x m median filter to have the value of the median intensity of the pixels
in the neighborhood
Median Filter (Nonlinear)
MAX and MIN filter
• Maximum and minimum filters attribute to each pixel in an image a new value equal to the maximum or
minimum value in a neighbourhood around that pixel.

• The neighbourhood stands for the shape of the filter.

• Maximum and minimum filters have been used in contrast enhancement.

• MAX filter is useful for finding the brightest points in an image or for eroding dark regions adjacent to bright
areas. Also, because pepper noise has very low values, it is reduced by this filter as a result of the max
selection process in the subimage area S xy.

• MIN filter is useful for finding the darkest points in an image or for eroding light regions adjacent to dark
areas. Also, it reduces salt noise as a result of the min operation
Applications of image restoration
• The digital image-restoration technique is widely used in the following fields of image processing:
1. Astronomical imaging
2. Medical imaging
3. Printing industry
4. Defense applications

Medical Imaging
• X-rays, mammograms and digital angiographic images are often corrupted by Poisson noise.
• Additive noise is common in magnetic resonance imaging.
• These noises should be removed for proper diagnostics of diseases.
• This can be accomplished through digital image-restoration techniques.
Astronomical Imaging
• The technical challenges faced by astronomers in gathering and interpreting images and spectra forced
the image-processing community to develop efficient image-restoration algorithms.
• Some of the problems include imaging through turbulent atmosphere, imaging of faint objects with low
photon counts and thus noisy data and the loss of phase information in radio interferometry.
• Astronomical image degradations are characterised by Poisson noise. Poisson noise is signal dependent.
• Another type of noise that is common is Gaussian noise which is due to electronic components in the
imaging system and broadcast transmissioning effect.
• Astronomical images are degraded by motion blur which is due to slow camera shutter speeds relative to
rapid spacecraft motion.
• To minimise the noise as well as degradation, astronomical imaging is one of the primary applications of
digital image restoration.
Printing Applications
• Printing applications often require the use of restoration to ensure that halftone
reproductions of continuous images are of high quality.
• Image restoration can improve the quality of continuous images generated from
halftone images

Defense Applications
• The image obtained from guided missiles may be degraded due to the effects of
pressure differences around a camera mounted on the missile.
• Proper image-restoration technique is a must to restore these images
Noise models
Gaussian
• Gaussian noise models are used frequently in practice
• Due to its mathematical tractability in both the spatial and
frequency domains,
Uniform noise
Additive noise
• Let f [m, n] be the original image, f ′[m, n] be the noise digitised version and η[m, n] be
the noise function which returns random values coming from an arbitrary distribution.
Then the additive noise is given by the equation

f ′[m, n] = f [m, n] + η[m, n]

• Additive noise is independent of the pixel values in the original image. Typically, η[m,
n] is symmetric about zero.

• This has the effect of not altering the average brightness of the image.

• Additive noise is a good model for the thermal noise within photo-electronic sensors.
Salt-Pepper noise

You might also like