Register No.

SRM Institute of Science and Technology

College of Engineering and Technology
SET - C
School of Computing
(Common to all branches)
Academic Year: 2023-24 (ODD)

Test: CLA-T1 Date: 20-2-2024

Course Code & Title: 21CSE251T DIGITAL IMAGE PROCESSING Duration: 100 minutes
Year & Sem: II Year / IV Sem Max. Marks: 50

Course Articulation Matrix: (to be placed)

Course
S.No. PO1 PO2 PO3 PO4 PO5 PO6 PO7 PO8 PO9 PO10 PO11 PO12 PSO1 PSO2 PSO3
Outcome

1 CO1 3 2 - - - - - - - - - - - 2 -
2 CO2 3 2 - 1 - - - - - - - - - 2 -
3 CO3 3 - 2 - 2 - - - - 1 - - - 2 -
4 CO4 3 2 - 1 - - - - - - - - - 2 -

5 CO5 3 - 2 1 2 - - - - 1 - - - 2 -

Part – B
(4 x 5 = 20 Marks)
Answer All 4 Questions
21a Consider that you are leading a training session for a 5 L2 1 1 1.3.1
team of junior engineers who are new to the field of
Digital Image Processing (DIP). Your goal is to
familiarize them with the basic steps involved in DIP and
how they contribute to various applications. How would
you structure the discussion and engage the team to
ensure they understand the fundamental steps in DIP and
their significance in practical scenarios?
Answer:
To effectively lead a training session for junior engineers
new to Digital Image Processing (DIP), it's important to
structure the discussion in a way that is engaging,
informative, and conducive to learning. Here's a
suggested approach:

Introduction to DIP: Start by providing a brief overview

of what DIP entails and its importance in various fields
such as medicine, security, entertainment, etc. Highlight
its role in enhancing and analyzing digital images for
better understanding and decision-making.
Basic Steps in DIP:
Image Acquisition: Explain the process of capturing
digital images using devices such as cameras or scanners.
Discuss the importance of image resolution, colour depth,
and other parameters in image quality.
Preprocessing: Introduce preprocessing techniques such
as noise reduction, image enhancement, and image
restoration. Emphasize how these techniques improve the
quality and usability of digital images.
Image Segmentation: Explain the process of dividing an
 image into meaningful regions or segments. Discuss
techniques like thresholding, edge detection, and region
growing, and their applications in object detection and
recognition.
Feature Extraction: Describe how features such as colour,
texture, shape, etc., are extracted from images to
represent their content. Discuss the significance of
feature extraction in tasks like pattern recognition and
image classification.
Image Analysis and Interpretation: Discuss methods for
analyzing and interpreting image data, such as
morphological operations, filtering, and mathematical
modelling. Highlight their role in extracting useful
information from images for decision-making.
Image Understanding: Introduce higher-level concepts
such as object tracking, scene understanding, and image
understanding. Discuss how these concepts integrate
various DIP techniques to achieve advanced tasks like
autonomous navigation and intelligent surveillance.
Practical Applications: Provide real-world examples and
case studies to illustrate the practical significance of DIP.
Show how DIP techniques are used in medical imaging
for diagnosis, in satellite imagery for environmental
monitoring, in security systems for facial recognition,
etc. Encourage discussion and questions from the team to
reinforce understanding.
OR
21b Provide examples of successful applications or products 5 L2 1 1 1.3.1
that leverage DIP components effectively to deliver
valuable features and functionalities to users. How do
these examples demonstrate the practical relevance and
impact of digital image processing in today's technology-
driven world?
Digital Image Processing (DIP) components are essential
in various applications and products across different
industries, enabling them to deliver valuable features and
functionalities to users. Here are some examples of
successful applications and products that leverage DIP
effectively:

Medical Imaging Systems:

MRI and CT Scanners: These systems utilize DIP

algorithms for image enhancement, noise reduction, and
segmentation to assist radiologists in diagnosing diseases
accurately.
Digital Pathology: DIP techniques are applied to
digitized tissue samples for tasks such as cancer
detection, tissue segmentation, and quantification of
cellular features, aiding pathologists in diagnosis and
treatment planning.
Autonomous Vehicles:

Computer Vision Systems: Autonomous vehicles use

DIP algorithms for object detection, recognition, and
tracking to perceive their environment and make
informed decisions. Examples include pedestrian
detection, lane detection, and traffic sign recognition.
LiDAR Data Fusion: DIP techniques are used to process
 and fuse LiDAR data with camera images for robust
perception in various weather and lighting conditions.
Smartphones and Cameras:
Face Recognition: DIP is utilized for facial recognition in
smartphone authentication systems, photo tagging, and
augmented reality applications.
Image Filters and Effects: Popular smartphone camera
apps apply DIP algorithms for real-time image
enhancement, beautification, and artistic effects.
Security and Surveillance Systems:
Video Analytics: Security cameras employ DIP
techniques such as motion detection, object tracking, and
behavior analysis for real-time monitoring and threat
detection.
Biometric Systems: DIP is used for fingerprint
recognition, iris recognition, and gait analysis in access
control and surveillance applications.
Satellite and Aerial Imaging:
Remote Sensing: Satellite and aerial imaging platforms
utilize DIP algorithms for land cover classification,
environmental monitoring, and disaster management.
Change Detection: DIP techniques are applied to detect
and analyze changes in Earth's surface over time, aiding
in urban planning, agriculture, and environmental studies.
These examples demonstrate the practical relevance and
impact of digital image processing in today's technology-
driven world by showcasing how DIP enables
advancements in various fields, including healthcare,
transportation, photography, security, and environmental
monitoring. DIP algorithms enhance the capabilities of
products and applications, enabling them to extract
meaningful information from visual data, make
autonomous decisions, improve user experiences, and
address real-world challenges efficiently.
22a Imagine you're leading a workshop for a group of 5 L2 1 1 1.3.1
aspiring digital artists who are eager to learn about image
processing techniques to enhance their artwork. To start
the workshop, you decide to explain the fundamental
concepts of pixels and the relationships between pixels in
digital images. How would you convey these concepts in
a way that engages the participants and helps them
understand the importance of pixels in image processing?

To begin the workshop for aspiring digital artists, it's

crucial to convey the fundamental concepts of pixels and
their importance in image processing in an engaging and
accessible manner. Here's how I would structure the
explanation:
Introduction to Pixels:
Start by defining what a pixel is: a fundamental unit of a
digital image, typically a small square or dot,
representing a single point in an image.
Explain that pixels are the building blocks of digital
images, and the resolution of an image refers to the
number of pixels it contains. Higher resolution means
more pixels, resulting in finer detail and clarity.
Visual Demonstration:
Use visual aids like images or graphics to illustrate the
concept of pixels. Show a zoomed-in view of an image to
 reveal individual pixels, emphasizing how they come
together to form the entire image.
Alternatively, use interactive tools or software that allow
participants to manipulate pixels directly, providing a
hands-on experience.
Understanding Pixel Relationships:
Discuss the relationships between neighboring pixels and
how they contribute to the overall appearance of an
image.
Explain concepts such as pixel intensity (grayscale) or
color (RGB), and how variations in intensity or color
among neighboring pixels create shapes, patterns, and
textures in an image.
Importance of Pixels in Image Processing:
Emphasize the significance of pixels in image processing
for digital artists. Pixels serve as the raw material that
artists manipulate to create, edit, and enhance their
artwork.
Highlight how understanding pixels enables artists to
control aspects like brightness, contrast, color balance,
and sharpness in their digital creations.
Discuss the role of pixels in special effects, image
manipulation, and digital composition, showcasing
examples from various art forms such as photography,
graphic design, and digital painting.
Interactive Exercises:
Engage participants in hands-on exercises that
demonstrate the impact of pixel manipulation on image
appearance. For example, ask them to adjust the
brightness or contrast of an image using image editing
software.
Encourage experimentation and creativity by challenging
participants to create their own digital artwork using
basic pixel manipulation techniques.
OR
22b Presume that you're leading a training session for a team 4 L2 1 1 1.3.1
of software engineers who are developing an image
processing module for a new photography app. To ensure
they understand the fundamentals, you decide to explain
the concepts of image sampling and quantization using
real-world scenarios. How would you break down these
concepts and engage the team to ensure they grasp the
importance of image sampling and quantization in image
processing?
Introduction to Image Sampling:
Definition: Image sampling refers to the process of
converting a continuous image into a discrete
representation by selecting a finite set of points (pixels)
from the continuous image space.
Real-world analogy: Imagine taking a photograph with a
digital camera. The camera's sensor samples the
continuous light in the scene and converts it into discrete
pixels, which collectively form the digital image.
Importance of Image Sampling:
Preservation of detail: Proper sampling ensures that
important details in the image are accurately captured
and represented by pixels.
Avoiding aliasing: Insufficient sampling can lead to
aliasing artifacts, where high-frequency details appear
 distorted or misrepresented in the sampled image.
Real-world scenario: Show examples of poorly sampled
images where fine details are lost or distorted due to
inadequate sampling.
Introduction to Image Quantization:
Definition: Image quantization involves assigning
discrete intensity levels (e.g., grayscale or color values)
to the sampled pixels obtained through sampling.
Real-world analogy: Think of quantization as the process
of choosing colors from a limited palette to represent the
colors in a painting or photograph.
Importance of Image Quantization:
Color representation: Quantization determines the
number of distinct colors or shades that can be
represented in the image.
Storage efficiency: Quantization reduces the amount of
data needed to represent the image, making it more
storage-efficient.
Real-world scenario: Provide examples of different
quantization levels and their impact on image quality and
file size, demonstrating how higher quantization levels
can lead to loss of color fidelity but smaller file sizes.
Interactive Demonstration:
Use interactive tools or software to demonstrate the
effects of sampling and quantization on images in real-
time.
Encourage team members to experiment with different
sampling rates and quantization levels to observe their
effects on image quality and file size.
23a Describe how to incorporate image sensing and 5 L2 1 1 1.3.1
acquisition into image processing to achieve accurate and
efficient quality image.
Incorporating image sensing and acquisition into image
processing to achieve accurate and efficient quality
images involves several steps and considerations. Here's
a general overview of the process:
Image Sensing and Acquisition:
Choose an appropriate imaging device such as a camera
or scanner based on your specific requirements
(resolution, sensitivity, speed, etc.).
Ensure proper calibration of the imaging device to
maintain consistency and accuracy in image acquisition.
Pay attention to factors like lighting conditions, exposure
settings, and focus to capture high-quality images.
Image Preprocessing:
Remove any noise or artifacts introduced during image
acquisition using techniques such as denoising filters or
image enhancement algorithms.
Correct for any distortions or imperfections in the
captured image, such as lens distortion or geometric
transformations.
Normalize the image to ensure consistent brightness and
contrast across different images.
Feature Extraction:
Identify relevant features in the image using techniques
like edge detection, corner detection, or blob detection.
Extract features that are crucial for the specific image
processing task at hand, such as object recognition,
segmentation, or classification.
 Image Processing Algorithms:
Choose appropriate image processing algorithms based
on the desired outcome of the analysis.
Implement algorithms for tasks such as image
segmentation, object detection, pattern recognition, or
image classification.
Optimize algorithms for efficiency and accuracy,
considering factors like computational complexity and
memory usage.
Post-Processing:
Refine the processed image using techniques like
morphological operations, filtering, or image blending.
Fine-tune parameters and adjust settings to improve the
overall quality and visual appearance of the processed
image.
Evaluation and Validation:
Assess the performance of the image processing pipeline
using quantitative metrics such as accuracy, precision,
recall, or F1-score.
Validate the results against ground truth data or human
annotations to ensure the reliability and correctness of the
processed images.
Iterate on the image processing pipeline as needed to
address any issues or improve performance.
Integration and Deployment:
Integrate the image processing pipeline into the larger
system or application where it will be used.
Ensure compatibility with other components and systems,
and consider factors like real-time processing
requirements or scalability.
Deploy the system in the target environment, monitoring
performance and making adjustments as necessary.
OR
23b As a digital image processing expert tasked with 5 L2 1 1 1.3.1
improving the quality of poorly lit images for a
photography studio's client, which techniques would you
employ for (i) image enhancement and (ii) image
restoration? Explain how each technique would
contribute to improving the overall quality of the images
in this scenario.
(i) Image Enhancement Techniques:

Histogram Equalization: This technique can help

improve the overall contrast and brightness of poorly lit
images by redistributing pixel intensities. It enhances the
visibility of details in both dark and bright areas, making
the image more visually appealing.

Contrast Stretching: By stretching the range of pixel

intensities in the image, contrast stretching can enhance
the perceptual contrast, making the details in poorly lit
areas more visible without significantly affecting the
well-lit areas.

Local Adaptive Enhancement: Techniques such as

adaptive histogram equalization or adaptive contrast
enhancement can be used to enhance local regions of the
image separately. This helps in preserving details and
avoiding over-enhancement in well-lit areas while
 improving the visibility of details in poorly lit regions.

Noise Reduction: Since poorly lit images often suffer

from increased noise levels, applying noise reduction
techniques such as median filtering or Gaussian
smoothing can help improve the overall quality by
reducing the distracting effects of noise.

(ii) Image Restoration Techniques:

Deconvolution: In poorly lit images, blurring may be a

significant issue due to low light conditions or camera
motion. Deconvolution techniques can help recover
sharpness and detail by estimating and removing the blur
introduced during image acquisition.

Retinex-Based Methods: Retinex algorithms aim to

restore the image's color and brightness to their true
values by separating illumination and reflectance
components. This can help in restoring the natural
appearance of the scene, especially in poorly lit
conditions where colors may appear washed out.

Super-Resolution: Super-resolution techniques can be

used to enhance the spatial resolution of poorly lit
images, thereby improving the level of detail and
sharpness. This is particularly useful when the original
image suffers from low-light-induced blur or pixelation.

Deep Learning-Based Restoration: Utilizing

convolutional neural networks (CNNs) trained
specifically for image restoration tasks can be highly
effective in improving the quality of poorly lit images.
These networks can learn complex mappings from input
to output images, enabling them to restore details and
enhance overall image quality.
24 a In the context of improving the photo editing tool's 5 L2 2 1 1.3.1
performance, discuss how understanding spatial and
frequency domains can inform the development of new
features or algorithms. How might leveraging techniques
from both domains enhance tasks such as image
enhancement, noise reduction, or feature extraction
within the software?
Spatial Domain:
In the spatial domain, images are represented as two-
dimensional arrays of pixel values.
Basic operations such as brightness adjustment, contrast
enhancement, and color correction are performed directly
on the pixel values.
Spatial filters, such as convolution kernels, are applied
directly to the image to perform operations like blurring,
sharpening, or edge detection.
Understanding spatial relationships between neighboring
pixels is essential for tasks like noise reduction and
image smoothing.
Techniques such as histogram equalization can be used to
enhance the overall contrast and dynamic range of an
image.
Frequency Domain:
 In the frequency domain, images are represented as a
combination of different spatial frequencies.
This representation allows us to analyze the frequency
content of an image using techniques like the Fourier
Transform.
High-frequency components correspond to rapid changes
in pixel intensity, such as edges and textures, while low-
frequency components represent smoother areas.
Filters in the frequency domain can be used to selectively
remove or enhance specific frequency components of an
image.
Techniques like low-pass filtering are effective for noise
reduction, as they suppress high-frequency noise while
preserving important image features.
Leveraging techniques from both domains can lead to
more sophisticated and effective photo editing
algorithms:
Image Enhancement:
Combining spatial domain operations like contrast
adjustment with frequency domain analysis can lead to
more targeted enhancements.
For example, identifying and enhancing specific
frequency components corresponding to important image
features like textures or edges while preserving overall
image quality.
Noise Reduction:
Utilizing both spatial and frequency domain techniques
can result in more robust noise reduction algorithms.
Spatial domain filters like median or Gaussian filters can
be combined with frequency domain filtering to
effectively remove noise while preserving image details.
Feature Extraction:
By analyzing the frequency content of an image, it's
possible to extract important features such as edges,
textures, or shapes more accurately.
Techniques like the Gabor filter, which combines both
spatial and frequency domain characteristics, can be used
for feature extraction tasks.
OR
24 b Provide insights into the selection and customization of 5 L2 2 1 1.3.1
smoothing spatial filters for the mobile app. How would
you determine the appropriate filter parameters, such as
kernel size and weighting coefficients, to achieve optimal
results for different types of images and noise levels?
Understanding Image Characteristics:
Analyze the types of images typically processed by the
app. Are they mostly portraits, landscapes, or macro
shots? Different types of images may require different
smoothing techniques.
Consider the presence of noise in the images. Is it
primarily Gaussian noise, salt-and-pepper noise, or a
combination of both? The type of noise will influence the
choice of smoothing filter.
Selecting the Filter Type:
Determine which spatial filter best suits the task at hand.
Common options include:
Gaussian filter: Effective for removing Gaussian noise
and preserving image details.
Median filter: Particularly useful for removing salt-and-
 pepper noise while preserving edges.
Bilateral filter: Balances noise reduction with edge
preservation, making it suitable for a wide range of
images.
Choosing Kernel Size:
The kernel size determines the extent of smoothing
applied to the image. Larger kernels produce more
extensive smoothing but may blur fine details.
Consider the trade-off between noise reduction and
preservation of image details. Smaller kernels are better
at preserving details but may not adequately suppress
noise.
Experiment with different kernel sizes based on the level
of noise and the desired amount of smoothing. For
example, start with smaller kernel sizes for images with
less noise and larger kernel sizes for images with higher
noise levels.
Adjusting Weighting Coefficients:
Some filters, like the Gaussian filter, require weighting
coefficients to determine the influence of neighboring
pixels on the smoothing process.
The choice of weighting coefficients affects the shape
and characteristics of the filter's response.
Experiment with different coefficient values to achieve
the desired balance between noise reduction and
preservation of image details.
Consider pre-computed coefficient sets optimized for
various applications or provide users with the option to
adjust coefficients manually for more control.
Testing and Validation:
Implement the selected filter configurations into the
mobile app.
Test the filters on a diverse set of images with varying
noise levels and characteristics.
Solicit feedback from users to assess the effectiveness of
the smoothing filters and identify any areas for
improvement.
Consider implementing adaptive filtering techniques that
automatically adjust filter parameters based on image
content and noise characteristics for more robust
performance.
Part – C
(1 x 10 = 10 Mark)
25 a Considering the requirements of the design firm's 10 L3 2 1 1.3.1
software tool, discuss the advantages and limitations of
different color image models, such as RGB, CMY, HSV,
and Lab. How would you determine which color
model(s) to incorporate into the software to best meet the
needs of graphic designers. Also tell how do these
models facilitate tasks such as color correction, image
editing, and color matching in professional design
workflows?
RGB (Red, Green, Blue):
Advantages:
Widely used in digital imaging and display devices.
Represents colors directly as combinations of red, green,
and blue components, making it intuitive for digital
workflows.
Well-suited for tasks involving additive color mixing,
such as digital painting, color grading, and designing for
screen-based media.
Limitations:
Not perceptually uniform, meaning equal changes in
RGB values may not correspond to equal changes in
perceived color.
Limited in its ability to describe colors outside the gamut
of typical display devices.
CMY (Cyan, Magenta, Yellow):
Advantages:
Used in subtractive color mixing, making it suitable for
printing workflows where colors are layered on a white
substrate.
Complements the RGB model, as CMY can be used to
create the subtractive color primaries for printing.
Allows for easy understanding of color mixing in print
production.
Limitations:
Inaccuracies in color reproduction due to imperfect
subtractive primaries and limitations of printing
processes.
Not as intuitive for digital design workflows as RGB.
HSV (Hue, Saturation, Value):
Advantages:
Represents colors in terms of perceptually meaningful
attributes: hue, saturation, and value.
Intuitive for tasks involving color selection and
adjustment, as it separates the color information from its
brightness.
Facilitates tasks such as color correction and adjustment
of color attributes independently.
Limitations:
Non-linear relationships between HSV components and
RGB values, leading to potential inconsistencies.
Limited in its ability to precisely represent colors outside
the RGB gamut.
Lab (CIELAB):
Advantages:
Designed to be perceptually uniform, meaning equal
distances in Lab space correspond to equal perceptual
differences in color.
Separates color information from lightness and
chromaticity, providing more control over color
adjustments.
Suitable for color correction, image editing, and color
matching tasks where accurate and consistent color
reproduction is crucial.
Limitations:
Complex mathematical transformations required to
convert between Lab and RGB, potentially impacting
performance in real-time editing workflows.
Limited support in some software and hardware
environments compared to RGB and CMY.
To determine which color model(s) to incorporate into
the software to best meet the needs of graphic designers,
consider the following factors:
Workflow Requirements: Assess the primary tasks
performed by graphic designers and the color workflows
they employ, such as digital design, print production, or
 web design.
Color Accuracy: Determine the level of color accuracy
and consistency required for the intended output, whether
for digital displays or print media.
User Preference: Solicit feedback from graphic designers
to understand their preferences and familiarity with
different color models.
Software Capabilities: Consider the software's
capabilities in handling color models and conversions, as
well as compatibility with industry-standard formats and
workflows.
In professional design workflows, these color models
facilitate tasks as follows:
Color Correction: Lab color model is often preferred for
precise color correction due to its perceptual uniformity
and separation of color information from lightness.
Image Editing: RGB and HSV models are commonly
used for digital image editing, providing intuitive
controls for adjusting colors and attributes.
Color Matching: CMYK is essential for achieving
accurate color reproduction in print production, while
Lab may also be used for precise color matching across
different devices and media.
Or
25 b How would you utilize histogram equalization to 10 L3 2 1 1.3.1
improve the visual quality of given image f(x,y)? Walk
through the steps involved in performing histogram
equalization on this image, considering the provided
intensity distribution.

f(x,y) =

(Handwritten copy)

*Program Indicators are available separately for Computer Science and Engineering in AICTE
examination reforms policy.

Course Outcome (CO) and Bloom’s level (BL) Coverage in Questions

BL Coverage in %
BL1
20%20%
BL2
60% BL3

Approved by the Audit Professor/Course Coordinator

How would you utilize histogram equalization to improve the visual quality of given image f(x,y)?
Walk through the steps involved in performing histogram equalization on this image, considering the
provided intensity distribution.

f(x,y) =