1. Derive the DCT coefficients from the definition of DFT.

Ans: To derive the Discrete Cosine Transform (DCT) coefficients from the definition of
the Discrete Fourier Transform (DFT) in digital image processing, we first need to recall
the basic definitions of the DFT and the DCT.
➢ Discrete Fourier Transform (DFT)
The DFT of a 1D sequence x[n]x[n]x[n] of length NNN is defined as:

➢ Discrete Cosine Transform (DCT)

The DCT is a transform that operates similarly to the DFT but only involves cosines.
Specifically, the DCT of type-II for a sequence x[n] of length NNN is given by:

➢ Relating DCT to DFT

To see how the DCT is related to the DFT, we start by examining the DCT formula and
comparing it with the DFT.

2. Explain the image restoration/degradation model.

Ans: Image Degradation Model
The degradation model mathematically represents how an image gets corrupted.
Typically, an image g(x, y) is considered a degraded version of the original image f(x, y).
• HH is the degradation function (often a convolution process),
• η(x, y) represents the noise added to the image.
Common forms of degradation include:
• Blurring: Caused by motion or out-of-focus optics, represented by a point spread
function (PSF).
• Noise: Random variations in the image intensity due to electronic sensor noise,
atmospheric disturbance, etc.
Image Restoration Model

1
The goal of image restoration is to recover the original image f(x, y) from the degraded
image g(x, y). There are several techniques used for this purpose:
➢ Inverse Filtering: Directly inverts the degradation function.
➢ Wiener Filtering: Optimal approach minimizing the mean square error between
the restored and original image.
➢ Regularization Methods: Introduce additional constraints or prior knowledge to
stabilize the inversion process.
➢ Blind Deconvolution: Used when the degradation function HH is unknown.

3. Explain the various types of noise models discussed in image processing.

Ans: Noise is an unwanted variation in an image, and understanding its types is crucial
for effective image processing and restoration. Here are some common noise models in
digital image processing:
➢ Gaussian Noise:
o Characteristics: Also known as normal noise, it has a probability density
function (PDF) equal to that of the normal distribution.
o Cause: Often arises due to thermal noise from electronic components.
➢ Salt-and-Pepper Noise (Impulse Noise):
o Characteristics: Manifests as randomly occurring white and black pixels.
o Cause: Typically caused by bit errors during image transmission or
malfunctioning pixels in sensor cameras.
➢ Poisson Noise (Shot Noise):
o Characteristics: The noise magnitude is proportional to the signal intensity,
and it follows a Poisson distribution.
o Cause: Commonly arises in photon-limited imaging, such as astronomical
or medical imaging.
➢ Speckle Noise:
o Characteristics: Multiplies the original image with a noise term, creating
a granular texture.
o Cause: Frequently seen in coherent imaging systems such as laser, SAR
(Synthetic Aperture Radar), and ultrasound.
➢ Uniform Noise:

2
 o Characteristics: All values within a specified range are equally likely.
o Cause: Can occur due to quantization effects in analog-to-digital
conversion.
4. Differentiate image enhancement and image restoration.
Ans:

5. What do you mean by spatial filtering? Explain.

Ans: Spatial filtering is a technique used in digital image processing to enhance or
modify images by applying a filter directly to the pixel values. It involves convolution
operations in the spatial domain and is used for various purposes such as smoothing,
sharpening, and edge detection.
Types of Spatial Filters
➢ Linear Filters: These filters work by taking the weighted sum of the pixel values
in the neighbourhood. Most common image processing filters are linear.
➢ Non-linear Filters: Non-linear filters do not take a weighted sum of neighbouring
pixel values but apply other operations, such as selecting the median or mode of
the neighbouring pixels. These are often used for noise reduction.
Common Spatial Filtering Techniques
➢ Smoothing/Blurring Filters: These filters reduce the detail in an image, often
used for noise reduction or to remove high-frequency components.
➢ Sharpening Filters: Sharpening filters highlight high-frequency components in an
image, enhancing edges and fine details.

3
 ➢ Edge Detection Filters: Edge detection filters highlight areas where there are
significant intensity changes in an image, usually at object boundaries.
➢ Noise Reduction Filters: These filters aim to remove noise from an image.

6. Give the practical limitations in sampling and reconstruction of images.

Ans: When it comes to sampling and reconstructing images in digital image processing,
several practical limitations can affect the quality and accuracy of the final image. Here
are some key limitations to consider:
Practical Limitations
➢ Aliasing:
o Issue: Occurs when the sampling rate is too low to capture the variations in
the image, leading to distortion and artifacts.
o Solution: Use anti-aliasing filters and ensure the sampling rate meets the
Nyquist criterion.
➢ Quantization Errors:
o Issue: When continuous image values are mapped to discrete levels,
quantization errors can introduce noise and reduce image quality.
o Solution: Increase the bit depth to reduce quantization noise, though this
also increases storage requirements.
➢ Loss of High-Frequency Details:
o Issue: During sampling, fine details may be lost, especially if the image
contains high-frequency content.
o Solution: Apply techniques like super-resolution to enhance high-frequency
details after sampling.
➢ Interpolation Artifacts:
o Issue: During image reconstruction, artifacts such as blurring or ringing can
occur due to interpolation methods.
o Solution: Use advanced interpolation techniques like spline interpolation or
bicubic interpolation to minimize artifacts.
➢ Noise Amplification:
o Issue: Noise present in the image can be amplified during the sampling and
reconstruction process.
4
 o Solution: Implement noise reduction techniques like filtering or wavelet
denoising prior to sampling.
➢ Resolution Limitations:
o Issue: The physical resolution of sensors and displays limits the achievable
image resolution.
o Solution: Use higher-resolution sensors and displays, but this comes with
higher costs and increased data size.

7. Explain the components of image processing.

Ans: Digital image processing involves various components that work together to
enhance, analyse, and manipulate images. Here are the key components:
➢ Image Acquisition: This is the first step in the process, where an image is
captured and converted into a digital form. This typically involves: Sensors and
Cameras, Digitization
➢ Image Enhancement: This component focuses on improving the visual quality of
an image, making it more suitable for a specific task. Techniques include:
Contrast Adjustment, Filtering, Histogram Equalization
➢ Image Restoration: Aimed at recovering the original image from a degraded
version by reversing the effects of noise and blurring. Common methods are:
Inverse Filtering, Wiener Filtering
➢ Image Segmentation: This component involves partitioning an image into
meaningful regions for easier analysis. Techniques include: Thresholding, Edge
Detection, Region-Based Segmentation
➢ Image Representation and Description: Converting processed image data into a
form suitable for further analysis: Boundary Representation, Region
Representation
➢ Image Compression: Reducing the amount of data required to represent an
image, facilitating storage and transmission. Methods include: Lossy
Compression, Lossless Compression

8. What are the different types of image smoothing filters? Discuss their
disadvantages.
Ans: In digital image processing, image smoothing filters are used to reduce noise and
smooth out the intensity variations in images. Here are some commonly used smoothing
filters along with their disadvantages:
Types of Image Smoothing Filters

5
 ➢ Mean Filter: Averages the pixel values within a specified window around each
pixel.
o Disadvantages: Can blur edges and fine details.
➢ Gaussian Filter: Uses a Gaussian function to weigh the pixel values within a
window, giving more importance to the central pixels.
o Disadvantages: Computationally more intensive due to the Gaussian
kernel.
➢ Median Filter: Replaces each pixel value with the median value of the pixels
within a specified window.
o Disadvantages: Can result in loss of fine detail.
➢ Bilateral Filter: Preserves edges while smoothing by combining the intensity and
spatial closeness of pixels.
o Disadvantages: Computationally expensive.
➢ Non-Local Means Filter: Averages pixels with similar intensities, even if they are
far apart, to better preserve textures.
o Disadvantages: Very computationally intensive.

9. Explain RGB, CMY & HIS color models emphasising on their respective
applications & significance.
Ans:
RGB Color Model
Description:
Components: Red (R), Green (G), Blue (B).
Colors are created by combining light of these three primary colors. By adjusting the
intensity of each color, a wide spectrum of colors can be produced.
Applications: Used in monitors, televisions, and camera sensors where colors are
generated by emitting light.
Significance: The RGB model aligns well with human vision, making it intuitive for
most digital display technologies.

CMY Color Model

Description:
6
 Components: Cyan (C), Magenta (M), Yellow (Y).
Colors are created by subtracting light. Used in printing, where colors are created by
overlapping inks.
Applications: Used in color printing processes like inkjet and laser printers, as well as in
color printing presses. The CMY model is often extended to CMYK, where K stands for
Key (black) to improve depth and contrast.
Significance: Directly relates to the physical process of color printing, making it
essential for producing accurate print colors.
HIS Color Model
Description:
Components: Hue (H), Saturation (S), Intensity (I).
Represents colors in terms of their chromatic content and brightness, aligning more
closely with human perception of color.
Applications: Useful in tasks that require color segmentation, enhancement, and
manipulation based on human perception, such as face detection and object recognition.
Significance: Allows for intuitive adjustments of colors, making it easier to enhance
images based on human visual preferences.

10. Discuss the concept, structure & applications of image pyramids.

Ans: Image pyramids are a powerful and fundamental tool in digital image processing,
used to represent images at multiple resolutions.
Concept
An image pyramid is a collection of images - all originating from a single original image,
with each subsequent image being a reduced resolution representation.
Structure
There are two main types of image pyramids:
➢ Gaussian Pyramid: Useful for image compression, multi-scale image analysis,
and object detection.
➢ Laplacian Pyramid: Widely used in image compression, blending, and
reconstruction tasks.
Applications

7
 ➢ Image Compression: By representing images at multiple resolutions, image
pyramids can help in efficient compression algorithms, like those used in JPEG
encoding.
➢ Image Blending and Seam Carving: Used in creating seamless transitions
between images, such as in panorama stitching or object removal.
➢ Object Detection and Recognition: Helps in detecting objects at different scales
and orientations by analysing images at multiple resolutions.

11. Discuss about wavelet transforms in one & two dimensions.

Ans: Wavelet transforms are a powerful tool in digital image processing, used for
analysing and representing signals at multiple resolutions. They decompose a signal or
image into a set of basic functions called wavelets, which are localized in both time and
frequency.
One-Dimensional Wavelet Transform
Concept:
• Continuous Wavelet Transform (CWT): Provides a continuous representation of
a signal by scaling and translating a mother wavelet.
• Discrete Wavelet Transform (DWT): A more practical version that uses discrete
scaling and translation steps, often implemented via filter banks.
Structure:
• Decomposition: The signal is decomposed into approximation coefficients and
detail coefficients using a pair of filters.
• Reconstruction: The original signal can be reconstructed by inverting the
decomposition process.
Applications:
• Data Compression: Efficiently compressing data by retaining significant wavelet
coefficients.
• Feature Extraction: Extracting meaningful features for pattern recognition and
classification.
Two-Dimensional Wavelet Transform
Concept:
• Extends the wavelet transform to two dimensions, commonly used for image
processing.

8
 • Applies the wavelet transform separately along the rows and columns of an
image, resulting in a multi-resolution representation.
Structure:
• Decomposition: The image is decomposed into four sub-bands: approximation
(LL), horizontal details (LH), vertical details (HL), and diagonal details (HH).
• Reconstruction: The original image can be reconstructed by combining the sub-
bands and inverting the filters.
Applications
• Image Denoising: Removing noise from images while preserving edges and
details.
• Image Fusion: Combining images from different sources to create a single,
enhanced image.
Advantages: Provides both time and frequency localization, offering better analysis of
non-stationary signals and images.
Example: The simplest and most basic wavelet, often used for educational purposes.

12. What do you mean n by color model? Explain the various color models.
Ans: A color model is a mathematical representation of colors using a set of primary
colors and a methodology for combining them. These models are essential for
processing, storing, and displaying color information in digital systems.
Various Color Models in Digital Image Processing
1. RGB (Red, Green, Blue) Color Model:
o Components: Red, Green, Blue.
o Additive Color Model: Combines light of different colors to create new
colors. By varying the intensity of these three components, a wide spectrum
of colors can be produced.
o Applications: Used in digital displays like monitors, TVs, and camera
sensors. Common in image editing and graphic design.
o Significance: Aligns well with human vision and digital display
technologies.
2. CMY (Cyan, Magenta, Yellow) Color Model:
o Components: Cyan, Magenta, Yellow.

9
 o Subtractive Color Model: Subtracts light using coloured inks or dyes to
create new colors. This model is often extended to CMYK, where K stands
for Key (black).
o Applications: Used in color printing processes like inkjet and laser printers.
o Significance: Directly relates to the physical process of color printing,
essential for accurate color reproduction.
3. HIS (Hue, Saturation, Intensity) Color Model:
o Components: Hue, Saturation, Intensity.
o Perceptual Model: Represents colors in terms of how humans perceive
them.
o Applications: Useful in image processing tasks like color segmentation,
enhancement, and computer vision.
o Significance: Provides an intuitive way to manipulate colors based on
human visual perception.
4. HSV (Hue, Saturation, Value) Color Model:
o Components: Hue, Saturation, Value.
o Perceptual Model: Similar to HIS, but Value replaces Intensity.
o Applications: Common in graphics software for color selection and
adjustment.
o Significance: Separates chromatic content from brightness, simplifying
color adjustments.

13. Explain the various image compression methods with necessary block
diagrams.
Ans: Image compression techniques aim to reduce the amount of data required to
represent an image while preserving its visual quality. They can be broadly categorized
into lossless and lossy compression methods.
1. Lossless Compression: Lossless compression algorithms reduce file size without
losing any data, ensuring the original image can be perfectly reconstructed.
Methods:
➢ Run-Length Encoding (RLE): Replaces sequences of the same pixel value with a
single value and a count.

10
 ➢ Huffman Coding: Uses variable-length codes for different pixel values based on
their frequencies.
➢ Lempel-Ziv-Welch (LZW): Builds a dictionary of commonly occurring
sequences and replaces them with shorter codes.
2. Lossy Compression: Lossy compression algorithms reduce file size by discarding
some of the image data, which may result in a loss of quality.
Methods:
➢ Discrete Cosine Transform (DCT): Transforms the image into the frequency
domain, quantizes the coefficients, and encodes them.
➢ JPEG Compression: Combines DCT, quantization, and Huffman coding.
➢ Wavelet-Based Compression: Uses wavelet transforms to decompose the image
into different frequency sub-bands and then quantizes and encodes them.
Applications
• Lossless Compression: Used in medical imaging, technical drawings, and
archival storage where loss of data is unacceptable.
• Lossy Compression: Used in everyday applications like web images, digital
photography, and streaming videos where some loss of quality is acceptable to
achieve higher compression ratios.

14. Describe the fundamental steps in image processing.

Ans: Digital image processing involves several fundamental steps that transform raw
image data into useful information or enhance the image quality. Here's a concise
overview of these key steps:
➢ Image Acquisition
Description:
• The first step involves capturing and converting an image into a digital format.
• Tools: Cameras, scanners, and sensors.
➢ Image Preprocessing
Description:
• Prepares the image for further analysis by improving its quality.
• Techniques:
o Noise Reduction: Using filters (e.g., Gaussian, median).
o Contrast Enhancement: Adjusting brightness and contrast.
11
 ➢ Image Enhancement
Description:
• Enhances the visual appearance of the image for better interpretation.
• Techniques:
o Histogram Equalization: Spreads out intensity values to improve contrast.
o Edge Enhancement: Highlights edges using filters like the Sobel filter.
➢ Image Restoration
Description:
• Aims to recover the original image from a degraded version.
• Techniques:
o Inverse Filtering: Reverses known degradation.
o Wiener Filtering: Minimizes error and reduces noise.
➢ Image Segmentation
Description:
• Divides an image into meaningful regions or objects.
• Techniques:
o Thresholding: Separates objects from the background.
o Edge Detection: Finds object boundaries using techniques like Canny edge
detection.
➢ Image Representation and Description
Description:
• Converts segmented regions into a form suitable for analysis.
• Techniques:
o Boundary Representation: Describes the shape using contours.
o Region Representation: Uses properties like texture and color.
➢ Image Compression
Description:
• Reduces the amount of data required to represent the image.
• Techniques:
12
 o Lossy Compression: Reduces file size by discarding some data (e.g.,
JPEG).
o Lossless Compression: Preserves all data (e.g., PNG).
➢ Image Recognition
Description:
• Identifies and classifies objects within an image.
• Techniques:
o Pattern Recognition: Uses feature extraction and classification algorithms.
o Machine Learning: Trains models to recognize patterns and objects.
➢ Color Image Processing
Description:
• Handles color images and processes them based on color information.
• Techniques:
o Color Space Conversion: Changes the representation of color (e.g., RGB to
HSV).
o Color Correction: Adjusts color balance and intensity.
➢ Morphological Processing
Description:
• Deals with the shape and structure of objects.
• Techniques:
o Erosion and Dilation: Shrinks or expands objects.
o Opening and Closing: Removes small objects and smooths boundaries.

15. Explain smoothing & sharpening filters in the context of spatial filtering.
Ans: Spatial filtering in digital image processing involves using a filter (or kernel) that
moves across an image to perform operations like smoothing and sharpening. Let's
explore both types of filters:
Smoothing Filters
Smoothing filters are designed to reduce noise and smooth out variations in an image.
They work by averaging the pixel values within a neighbourhood, thereby reducing
sharp transitions in intensity.
13
 ➢ Mean Filter:
o Description: Calculates the average of the pixel values within the kernel
and assigns this average to the central pixel.
o Original Image → Mean Filter → Smoothed Image
o Effect: Blurs the image and reduces noise, but can also blur edges and fine
details.
➢ Gaussian Filter:
o Description: Uses a Gaussian function to weight the pixels within the
kernel, giving more importance to the central pixels.
o Original Image → Gaussian Filter → Smoothed Image
o Effect: Provides a smoother and more natural blurring effect compared to
the mean filter, with reduced edge blurring.
➢ Median Filter:
o Description: Replaces each pixel value with the median value of the pixels
within the kernel.
o Original Image → Median Filter → Smoothed Image
o Effect: Effective at reducing salt-and-pepper noise while preserving edges
better than mean and Gaussian filters.
Sharpening Filters
Sharpening filters enhance edges and fine details in an image by emphasizing high-
frequency components.
➢ Laplacian Filter:
o Description: Uses the second derivative of the image to highlight regions of
rapid intensity change.
o Original Image → Laplacian Filter → Sharpened Image
o Effect: Emphasizes edges, but can also amplify noise.
➢ Sobel Filter:
o Description: Computes the gradient magnitude of the image using two 3x3
kernels for horizontal and vertical gradients.
o Original Image → Sobel Filter → Edge-Enhanced Image
o Effect: Highlights edges in specific directions, useful for edge detection.
➢ Unsharp Masking:
14
 o Description: Enhances edges by subtracting a blurred version of the image
from the original image.
o Original Image → Gaussian Blur → Subtract from Original → Sharpened
Image
o Effect: Enhances edges without significantly amplifying noise, commonly
used in photo editing.
Practical Applications
• Smoothing Filters: Used for noise reduction, background smoothing, and pre-
processing before further analysis.
• Sharpening Filters: Employed in edge detection, feature extraction, and image
enhancement for better visual quality.

16. What is noise? Differentiate between Gaussian noise & Impulsive noise?
Ans: Noise in digital image processing refers to random variations in pixel values that
can degrade the quality of an image. It often arises from various sources like electronic
sensor errors, environmental conditions, or transmission errors. Noise can obscure
important image details, making it challenging to analyze or interpret the image
accurately.
Gaussian Noise
Characteristics:
• Distribution: Gaussian noise, also known as normal noise, follows a normal
distribution with a bell-shaped probability density function.
• Parameters: Defined by its mean (average value) and standard deviation (spread).
• Appearance: Appears as grainy texture evenly distributed across the image.
Sources:
• Arises from thermal vibrations in electronic circuits and sensors.
• Common in low-light conditions where sensor noise is more prominent.
Effects:
• Can blur fine details.
• Reduces overall image clarity.
Impulsive Noise
Characteristics:
15
 • Distribution: Impulsive noise, also known as salt-and-pepper noise, introduces
random occurrences of black and white pixels.
• Appearance: Manifests as sparsely scattered white (salt) and black (pepper) pixels
against the original image.
Mathematical Model: Typically modelled as a random process where pixel values are
replaced by maximum or minimum intensity values with certain probabilities.
Sources:
• Caused by errors in data transmission or malfunctioning pixels in sensor arrays.
• Often arises in digital communication systems and faulty hardware.
Effects:
• Creates distinct spots or specks that can be visually disruptive.
• Can significantly affect image analysis and processing tasks.
Key Differences
• Distribution: Gaussian noise has a continuous distribution affecting all pixel
values, while impulsive noise affects specific pixels with high-intensity values.
• Appearance: Gaussian noise appears as a smooth, grainy texture, whereas
impulsive noise appears as distinct black and white specks.

16