Define high boost filtering and unsharp masking in image processing (3 marks)

Explain the basic image degradation / restoration model (3 marks)

Why ringing effect is produced in images while undergoing ideal low pass filtering (6
marks)

Cause of Ringing Effect

●​ The abrupt transition in the frequency domain causes oscillations in the spatial
domain when the inverse Fourier transform is applied.​

●​ This is due to the Gibbs phenomenon, which occurs whenever a discontinuity is

present in a signal or its spectrum.​

●​ As a result, overshoots and undershoots appear near edges in the restored image,
giving a wavy or ringing appearance.​

●​ The effect is more pronounced near sharp edges or high-contrast boundaries.​

Mathematically:​
The spatial response (impulse response) of an ILPF is a sinc function:

h(x,y) = sinc (x,y)

The oscillations of the sinc function are what cause the ringing effect

Define illuminance reflectance model of an image. Describe how frequency domain image
enhancement is achieved with image model
Definition of Illumination–Reflectance Model (3 Marks)

An image can be modeled as the product of illumination and reflectance components:

f(x,y)=i(x,y)⋅r(x,y)

Where:

●​ f(x,y) → Observed image intensity

●​ i(x,y) → Illumination component (amount of light falling on the scene)
●​ r(x,y) → Reflectance component (amount of light reflected by the object surface)​

2. Explanation:

●​ Illumination varies slowly and represents the low-frequency components of the

image.
●​ Reflectance varies rapidly and corresponds to high-frequency details such as edges
and textures.
●​ Most enhancement techniques aim to separate or modify these two components to
improve visibility.​

Frequency Domain Image Enhancement (3 Marks)

In the frequency domain, the image is represented by its Fourier Transform:

In the frequency domain, the image is represented by its Fourier Transform:

F(u,v)=F{f(x,y)}

Enhancement is performed by applying a filter function H(u,v)H(u, v)H(u,v):

G(u,v)=H(u,v)⋅F(u,v)

and the enhanced image is obtained by inverse transform:

g(x,y)=F−1{G(u,v)}

4. Types of Frequency Domain Filters:

●​ Low Pass Filter (LPF): Smoothens image, removes noise.

●​ High Pass Filter (HPF): Enhances edges and fine details (reflectance).
 ●​ High-Boost Filter: Combines LPF and HPF effects for controlled sharpening.

Differentiate between histogram specification and histogram equalization

Feature Histogram Equalization Histogram Specification

(Matching)

Definition A process that automatically A process that transforms the

redistributes the intensity values of intensity distribution of an image so
an image so that the histogram that its histogram matches a
becomes approximately uniform. desired (specified) histogram.

Objective To enhance overall contrast of the To obtain a specific intensity

image. distribution or visual appearance.

Input No reference histogram required. Requires a reference or target

Requirement histogram to match.

Control over Output contrast enhancement is Output appearance is controlled

Output automatic and unpredictable. and predictable based on the
reference histogram.

Applications General contrast improvement. Matching images from different

sources or lighting conditions.

Define inverse filtering for image restoration. What are its disadvantages (4 marks)
Disadvantages (2 Marks):

1.​ Highly sensitive to noise – Even small amounts of noise get amplified when H(u,v)
has small values.​

2.​ Instability problem – If H(u,v)=0 or near zero, division causes large distortions.​

3.​ Requires exact knowledge of H(u,v) – Inaccurate estimation of degradation leads to

poor restoration.​

4.​ Not suitable for real-world images – Since noise and imperfect models are common,
the results are often unsatisfactory.

If all the pixels in an image is shuffled will there be any changes in the histogram ?
Justify your answer (2 marks)

●​ The histogram represents only the frequency of occurrence of each gray level (i.e.,
how many pixels have a particular intensity value).
●​ ​
Shuffling pixels changes their spatial positions, but not their intensity values.​

●​ Hence, the gray-level distribution remains the same, and the histogram remains
unchanged.

​
What is a Wiener filter Explain (4 mark)
3. Explanation (1 Mark):

●​ The filter balances inverse filtering and noise smoothing.​

●​ When noise is very small, the Wiener filter behaves like an inverse filter.​

●​ When noise is large, it smoothens the image to reduce noise effects.​

4. Advantages / Key Point (1 Mark):

●​ Reduces both blur and noise simultaneously.​

●​ Provides an optimal restoration in the sense of minimum mean square error

(MMSE).

Draw and explain the steps involved in homomorphic filtering

Define the fundamentals of spatial filtering (3 marks)
Differentiate smoothing and sharpening techniques (3 marks)

Feature Smoothing Sharpening

Purpose Reduces noise and minor variations Enhances edges and fine details in an
in intensity; used for blurring. image.

Operation Averages or smooths pixel values Highlights intensity transitions using

using filters like mean or Gaussian filters like Laplacian or high-pass filter.
filter.

Effect on Makes the image less sharp and Increases contrast around edges,
Image removes noise. making the image appear clearer.

Briefly discuss 1. Negative 2. Power law 3. Contrast stretching (6 marks)

1. Negative Transformation (2 marks)

●​ The negative transformation produces an image that is the photographic negative of

the original.​
 ●​ It is defined as:​
s=L−1−r​
where rrr = input intensity, s = output intensity, and L = maximum gray level (e.g., 256
for 8-bit images).​

●​ Effect: Dark areas become light and light areas become dark — useful for enhancing
white or gray details embedded in dark regions.

Discuss the restoration of images degraded by additive noise, using any three order
statistics filter

Three Common Order-Statistics Filters

1. Median Filter

●​ Replaces each pixel value with the median of the intensity values in its neighborhood.​

●​ Excellent for removing salt-and-pepper noise while preserving edges.​

●​ Example: For a 3×3 window, sort the 9 pixel values and choose the middle one (5th
value).

Effect: Removes isolated bright or dark spots effectively.

2. Max Filter

●​ Replaces each pixel value with the maximum value in its neighborhood.​
 ●​ Useful for removing pepper noise (dark spots) caused by low-intensity impulses.​

●​ Also enhances bright regions.

Effect: Brightens image regions and eliminates dark noise points.

3. Min Filter

●​ Replaces each pixel value with the minimum value in its neighborhood.​

●​ Useful for removing salt noise (bright spots) caused by high-intensity impulses.​

●​ Also darkens bright areas slightly.​

Effect: Removes bright noise points and smooths light areas.