Image Processing Suggestions

1: What is a Pixel?

A: In digital imaging, a pixel(or picture element) is the smallest item of

information in an image. Pixels are arranged in a 2-dimensional grid,
represented using squares. Each pixel is a sample of an original image,
where more samples typically provide more-accurate representations of the
original. The intensity of each pixel is variable; in color systems, each pixel
has typically three or four components such as red, green, and blue, or cyan,
magenta, yellow, and black.

The word pixel is based on a contraction of pix ("pictures") and el (for

"element").

2: What is image resolution?

A: The term resolution is often used as a pixel count in digital imaging. When
the pixel counts are referred to as resolution, the convention is to describe
the pixel resolution with the set of two numbers. The first number is the
number of pixel columns (width) and the second is the number of pixel rows
(height), for example as 640 by 480. Another popular convention is to cite
resolution as the total number of pixels in the image, typically given as
number of megapixels, which can be calculated by multiplying pixel columns
by pixel rows and dividing by one million. An image that is 2048 pixels in
width and 1536 pixels in height has a total of 2048×1536 = 3,145,728 pixels
or 3.1 megapixels. One could refer to it as 2048 by 1536 or a 3.1-megapixel
image. Other conventions include describing pixels per length unit or pixels
per area unit, such as pixels per inch or per square inch.

Below is an illustration of how the same image might appear at different pixel
resolutions.
As the megapixels of a camera increase so does the ability of a camera to
produce a larger image; a 5 megapixel camera is capable of capturing a larger
image than a 3 megapixel camera.

Larger monitor screens usually have higher screen resolution, measured in

pixels.
3: What is DPI / PPI?

A: DPI refers to dots per inch when using an ink based printer. It is a
measure of resolution or image quality. Typically, the higher the dpi count,
the better the print quality. This term is still used when discussing digital
image quality; however, this is not the correct term.

PPI describes the resolution, in pixels, of an image to be printed within a

specified space. For instance, a 100x100-pixel image that is printed in a 1-
inch square could be said to have 100 pixels per inch, regardless of the
printer's DPI capability. Used in this way, the measurement is only
meaningful when printing an image. Good quality photographs usually
require 300 pixels per inch when printed.
4: How is a digital image's file size determined?

A: Image file size-expressed as the number of bytes-increases with the

number of pixels composing an image and the colour depth of the pixels. The
greater the number of pixel rows and pixel columns, the greater the image
resolution, and the larger the file. Also, each pixel of an image increases in
size when its colour depth increases—an 8-bit pixel (1 byte) stores 256
colors, a 24-bit pixel (3 bytes) stores 16 million colors, the latter known as
truecolor. Image compression uses algorithms to decrease the size of a file.
High resolution cameras produce large image files, ranging from hundreds
of kilobytes to megabytes, per the camera's resolution and the image-storage
format capacity. High resolution digital cameras may record 12 megapixel
(1MP = 1,000,000 pixels / 1 million) images, or more, in truecolor. For
example, an image recorded by a 12 MP camera; since each pixel uses 3
bytes to record truecolor, the uncompressed image would occupy 36,000,000
bytes of memory—a great amount of digital storage for one image, given that
cameras must record and store many images to be practical. Faced with large
file sizes, image file formats with built in compression routines were
developed to store such large images.

5: What is file compression?

A: When discussing the quality of digital files based on file size, comparisons
should only be made based on uncompressed sizes. Compression
algorithms will modify each image differently depending on the subject
matter of the image. Therefore it is impossible to accurately compare the file
size of two digital images once they have been compressed.

There are two types of file compression, "lossy" and "lossless". Lossy
compression actually changes some of the original pixels and some details
are lost. The most common format of lossy compression is JPEG. While the
original JPEG image out of a digital camera is fine, every time the file is saved
again, detail is lost. If the same file is saved as a JPEG several times,
significant quality is lost and cannot be recovered. Valuable originals should
always be saved in a lossless format, like TIFF or PSD. TIFF files can be
edited and saved any number of times without loss of detail because the
compression does not alter any pixels. The trade off is that TIFF files do not
compress as well as JPG.

6: What are the various file formats an image can be saved as?

A: Image file formats are standardized means of organizing and storing

images. Image files are composed of either pixel or vector data that are
rasterized to pixels when displayed on a computer monitor. The pixels that
compose an image are ordered as a grid (columns and rows); each pixel
consists of numbers representing magnitudes of brightness and color.
Including proprietary types, there are hundreds of image file types. The
JPEG, PNG, and GIF formats are most often used to display images on the
Internet. Digital cameras typically save images in the JPEG format which is
a lossy format, meaning image compression takes place to save memory
space and maximize the number of files one can fit on a memory card or
hard drive. Other formats include TIFF, PSD, RAW, and BMP.
7: What size, in megapixels, should a digital file be in order to produce
a quality print at a given size, in inches?

A: Each colored box represents a certain number of megapixels. The

numbers along the top and left side are print dimensions in inches at 300ppi
(pixels per inch). Most books and magazines require 300ppi for photo quality.
For example, the chart shows that you can make a 5" x 7" photo quality print
from a 3 megapixel camera.

Inches @ 300ppi (numbers inside colored boxes are megapixels)

Notice that as the print size doubles, the megapixels required increases
geometrically. You can make nice 8" x 10" prints with a 6 or 8 megapixel
camera, but to make a true photo quality 16" x 20" print, you need between
24 and 30 megapixels. Don't be fooled by manufacturers' claims that say you
can make 16" x 20" prints from an 8 megapixel camera. While you
certainly can make a print that size, it will not be true photo quality at
300ppi. Through the use of image editing software like Photoshop, one can
"cheat" by adding pixels to an image to increase its size. Image clarity will
not be improved however, as all the new pixels will be created by averaging
the values of the original pixels.
8.What is the difference between image processing and computer
vision?

Image processing is the process of manipulating digital images through a

computer. This can include tasks such as resizing, cropping, or adding
filters to an image. Computer vision, on the other hand, is the process of
using computers to interpret and understand digital images. This can
involve tasks such as object recognition or facial recognition.

9.Can you give some examples of real-world applications that use

digital image processing techniques?

1. Medical Imaging:
o X-ray, CT scans, and MRI images are processed to enhance
details, detect anomalies, and aid in diagnosis.
o Ultrasound images benefit from noise reduction and feature
extraction techniques.
o PET scans and gamma-ray imaging also rely on digital image
processing1.
2. Computer Vision:
o Object detection and tracking in surveillance systems.
o Facial recognition for security and authentication.
o Automated inspection in manufacturing (e.g., detecting
defects on assembly lines).
o Gesture recognition for human-computer interaction2.
3. Remote Sensing:
o Satellite imagery is processed to monitor land use, vegetation,
weather patterns, and environmental changes.
o Agriculture: Assessing crop health, estimating yield, and
detecting pests.
o Geographical Information Systems (GIS): Mapping and
spatial analysis3.
4. Robotics and Automation:
o Robot vision: Robots use image processing to navigate,
recognize objects, and perform tasks.
o Industrial automation: Inspecting products, sorting items, and
quality control2.
5. Entertainment and Media:
o Video compression for streaming and storage.
o Special effects in movies and animations.
o Image enhancement for photography and video editing4.
6. Biometrics:
o Fingerprint recognition, iris scanning, and vein pattern
analysis.
o Voice recognition and signature verification2.
7. Non-Photorealistic Rendering (NPR):
o Creating artistic or stylized images from photographs.
 o Examples include oil painting filters, sketch effects,
and cartoonization1.
8. Lane Departure Warning Systems:
o In automotive safety, image processing detects lane markings
and alerts drivers if they deviate from their lane 1.
9. Microscope Imaging:
o Enhancing microscopic images for medical research, biology,
and material science 1.
10. Forensics and Security:
o Enhancing surveillance footage for criminal investigations.
o Document analysis for detecting forged signatures or altered
tex

10.How do you differentiate between a high-pass filter and a low-pass

filter?

1. Low-Pass Filter (LPF):

o Purpose: A low-pass filter allows low-frequency signals to pass
through while attenuating high-frequency signals.
o Function: It is commonly used for smoothing or reducing noise
in signals.
o Preservation:
 Low-frequency components are preserved.
 High-frequency components are attenuated.
o Circuit Components:
 Typically consists of a resistor followed by a capacitor.
o Applications:
 Used in audio processing, image processing,
communication systems, and biomedical signal
processing.
2. High-Pass Filter (HPF):
o Purpose: A high-pass filter allows high-frequency signals to
pass through while attenuating low-frequency signals.
o Function: It is used for sharpening or emphasizing high-
frequency details.
o Preservation:
 High-frequency components are preserved.
 Low-frequency components are attenuated.
o Circuit Components:
 Typically consists of a capacitor followed by a resistor.
o Applications:
 Used in audio processing, image enhancement, and noise
removal.

11.What’s the difference between edge detection and line detection?

1. Edge Detection:
 o Definition: Edge detection aims to identify boundaries or
transitions between different objects or regions within an image.
o Objective: Detecting areas of significant intensity
variation or gradients in the image.
o Characteristics:
 Edges correspond to changes in color, texture,
or intensity.
 Crucial for tasks like object recognition, image
segmentation, and feature extraction.
o Steps in Edge Detection:
 Intensity Gradient: Calculate the intensity gradient of
the image.
 Derivative Operators: Use operators (e.g., Sobel, Prewitt,
Scharr) to compute the gradient.
 Gradient Magnitude: Represent the strength of intensity
change.
 Gradient Direction: Indicates the edge orientation.
 Thresholding: Classify pixels as edge or non-edge based
on gradient magnitudes.
 Non-maximum Suppression: Obtain thin, well-localized
edges.
 Hysteresis (Optional): Connect weak edges to strong
edges.
 Edge Representation: Output a binary image or pixel
coordinates representing detected edges1.
2. Line Detection:
o Definition: Line detection focuses on finding line segments (and
sometimes other geometric figures like circular arcs) within an
image.
o Characteristics:
 Identifies linear structures (lines) in the image.
 Useful for tasks like road detection, building outlines,
and geometry recognition.
o Relation to Edge Detection:
 Line detection is often performed on the output of an edge
detector.
 It operates on the edges detected by edge detection
algorithms.
 While edges are local, lines are non-local features

12. What type of filters are used to reduce noise in an image?

There are a variety of filters that can be used to reduce noise in an image,
but the most common are median filters and Gaussian filters. Median
filters work by replacing each pixel in an image with the median value of
the surrounding pixels, while Gaussian filters work by blurring an image
and then sharpening it again.
13.What is the importance of using feature vectors when performing
image classification?

Feature vectors are important when performing image classification

because they provide a way to reduce the dimensionality of the data while
still retaining important information about the image. This can be helpful
in cases where the data is too high-dimensional to be processed efficiently,
or where there is a lot of noise in the data that can be filtered out by using
a lower-dimensional representation. Additionally, using feature vectors can
make it easier to compare different images to each other and to find
patterns in the data.

14. Can you explain the differences between RGB, HSV, CMYK, and
YCbCr?

RGB, or red-green-blue, is the most common color model used in digital

image processing. HSV, or hue-saturation-value, is another common color
model that is often used to more easily identify colors. CMYK, or cyan-
magenta-yellow-black, is a color model used in printing. YCbCr, or luma-
chrominance, is a color model used in digital video and image processing.

1. RGB (Red-Green-Blue):
o Purpose: RGB is an additive color model used for digital
images and displays.
o Components: It consists of three channels: Red, Green,
and Blue.
o Color Creation: All other colors are produced by varying the
proportional ratios of these three primary colors.
o Representation: 0 represents black, and as the value
increases, color intensity increases.
o Applications: Used in digital image processing, openCV, and
online logos.
o Color Combinations:
 Green(255) + Red(255) = Yellow
 Green(255) + Blue(255) = Cyan
 Red(255) + Blue(255) = Magenta
 Red(255) + Green(255) + Blue(255) = White 1.
2. CMYK (Cyan-Magenta-Yellow-Black):
o Purpose: CMYK is widely used in printers.
o Subtractive Model: It is a subtractive color model.
o Components: Consists of four channels: Cyan, Magenta,
Yellow, and Black (Key).
o Color Creation: Colors are subtracted from white (1) to create
different shades.
o Representation: Point (1, 1, 1) represents black, and (0, 0, 0)
represents white.
o Color Relationships:
 Cyan = Negative of Red
  Magenta = Negative of Green
 Yellow = Negative of Blue 1.
3. HSV (Hue-Saturation-Value):
o Components:
 Hue: Represents different colors in a circular range (0 to
360 degrees).
 Saturation: Describes the percentage of color (0 to 1).
 Value: Represents intensity (0 to 100%).
o Perception-Based: HSV models color the way humans perceive
it.
o Applications: Used in histogram equalization and converting
grayscale images to RGB.
o Color Representation: Visualized as a cone, with different hues
at different angles1.
4. YCbCr:
o Purpose: YCbCr is widely used in television broadcasting.
o Components:
 Y (Luminance): Represents brightness.
 Cb and Cr (Chrominance): Indicate blue and red
components relative to green.
o Color Space: Exploits human eye properties.
o Representation: Separates color information into luminance
and chrominance components

15.What do you understand about illumination invariance? Is it

possible to achieve it in practice?

Illumination invariance is the ability of an image processing algorithm to

produce consistent results regardless of the level of illumination in the
scene being captured. In other words, it should not matter if it is a bright
sunny day or a dark night – the algorithm should still be able to produce
the same results. In practice, it is often difficult to achieve perfect
illumination invariance, but it is possible to get close.

16. Can you explain what scale invariance is? Do all images have this
property?

Scale invariance is the ability of an image to maintain its appearance when

scaled up or down. This means that the features of the image will remain
the same, even if the size of the image changes. Not all images have this
property, but many do.

1. Scale Invariance:
o Definition: Scale invariance refers to the property of an
algorithm or representation that remains consistent even when
the scale (size) of an object or feature changes.
 o Why It Matters:
 Real-world objects exist at various scales. For instance, a
tree can appear both as a tiny sapling and as a large,
mature tree.
 Algorithms that exhibit scale invariance can detect and
recognize the same features regardless of their size.
 Achieving scale invariance is crucial for robustness in
computer vision tasks.
2. Scale Invariance in Images:
o SIFT (Scale-Invariant Feature Transform): One of the most
notable examples of scale invariance is the SIFT algorithm. It
was introduced by D. Lowe in 2004.
 Key Properties:
 Locality: SIFT features are local, making them
robust to occlusion and clutter.
 Distinctiveness: Individual features can be matched
to a large database of objects.
 Quantity: Many features can be generated even for
small objects.
 Efficiency: SIFT achieves close-to-real-time
performance.
 Extensibility: It can be extended to different feature
types.
 Steps in SIFT:
1. Scale-Space Peak Selection: Identifying potential
feature locations across different scales.
2. Keypoint Localization: Accurately locating feature
keypoints.
3. Orientation Assignment: Assigning orientations to
keypoints.
4. Keypoint Descriptor: Describing keypoints as high-
dimensional vectors.
o Gaussian Scale Space: Real-world objects exhibit multi-scale
behavior. A scale space attempts to replicate this concept in
digital images.
 Images are progressively blurred using the Gaussian blur
operator at different scales (octaves).
 Blurring simulates the effect of viewing objects at different
distances.
 The Difference of Gaussians (DoG) is then computed from
these blurred images.
3. Not All Images Have Perfect Scale Invariance:
o While algorithms like SIFT aim for scale invariance, real-world
images may not always exhibit perfect scale invariance.
o Some features may change significantly with scale (e.g., fine
textures, small details).
 o Achieving complete scale invariance remains a challenge, but
techniques like down-sampling and convolutional networks help
mitigate scale variations

1. Scale Invariance:
o Definition: Scale invariance refers to the property of an
algorithm or representation that remains consistent even when
the scale (size) of an object or feature changes.
o Why It Matters:
 Real-world objects exist at various scales. For instance, a
tree can appear both as a tiny sapling and as a large,
mature tree.
 Algorithms that exhibit scale invariance can detect and
recognize the same features regardless of their size.
 Achieving scale invariance is crucial for robustness in
computer vision tasks.
2. Scale Invariance in Images:
o SIFT (Scale-Invariant Feature Transform): One of the most
notable examples of scale invariance is the SIFT algorithm. It
was introduced by D. Lowe in 2004.
 Key Properties:
 Locality: SIFT features are local, making them
robust to occlusion and clutter.
 Distinctiveness: Individual features can be matched
to a large database of objects.
 Quantity: Many features can be generated even for
small objects.
 Efficiency: SIFT achieves close-to-real-time
performance.
 Extensibility: It can be extended to different feature
types.
 Steps in SIFT:
1. Scale-Space Peak Selection: Identifying potential
feature locations across different scales.
2. Keypoint Localization: Accurately locating feature
keypoints.
3. Orientation Assignment: Assigning orientations to
keypoints.
4. Keypoint Descriptor: Describing keypoints as high-
dimensional vectors.
o Gaussian Scale Space: Real-world objects exhibit multi-scale
behavior. A scale space attempts to replicate this concept in
digital images.
 Images are progressively blurred using the Gaussian blur
operator at different scales (octaves).
 Blurring simulates the effect of viewing objects at different
distances.
  The Difference of Gaussians (DoG) is then computed from
these blurred images.
3. Not All Images Have Perfect Scale Invariance:
o While algorithms like SIFT aim for scale invariance, real-world
images may not always exhibit perfect scale invariance.
o Some features may change significantly with scale (e.g., fine
textures, small details).
o Achieving complete scale invariance remains a challenge, but
techniques like down-sampling and convolutional networks help
mitigate scale variations

17. What happens if you apply multiple transforms to an image?

1.
o CSS Transforms:
 In CSS, you can use the transform property to apply
various transformations to HTML elements, such as
rotation, translation (movement), scaling, and skewing.
 When you apply multiple transforms, they are chained
together in a specific order.
 The order of transformation matters because each
transform operates relative to the previous one.
o Transform Order:
 Right-to-Left: When you specify multiple transforms in a
single line, they are applied from right to left.
 The last transform listed is applied first, followed by the
second-to-last, and so on.
2. Matrix Representation:
o Under the hood, transformations are represented
using matrices.
o You can construct a 4x4 transformation matrix that combines
multiple transformations (rotation, translation, scaling) and
apply it to the image.
o Matrix multiplication allows you to combine transformations in
any order.

18.What are Gabor Filters?

A Gabor filter, named after Dennis Gabor, is a linear filter used

for texture analysis in image processing. Let’s explore its key properties
and applications:
1. Purpose:
o The Gabor filter analyzes whether there is any
specific frequency content in an image within a localized
region around a point or region of interest.
o It focuses on both frequency (how rapidly intensity changes)
and orientation (direction of intensity change).
2. Mathematical Representation:
 o The impulse response of a 2D Gabor filter is defined by
a sinusoidal wave (a plane wave) multiplied by a Gaussian
function.
o It has both real and imaginary components representing
orthogonal directions.
o These components can be combined into a complex number or
used individually.
3. Applications:
o Texture Representation: Gabor filters are particularly
appropriate for representing and discriminating textures.
o Edge Detection: They can detect edges with specific
orientations.
o Feature Extraction: Useful for extracting relevant features
from images.
o Computer Vision: Applied in tasks like object
recognition and image segmentation.

19. Short notes: Image acquisition

The principal phenomenon at the origin of the acquisition of an image is
the electromagnetic spectrum. Images based on radiation from the
electromagnetic spectrum are the most familiar, especially images from
visible light, as photography. Other images based on the electromagnetic
spectrum include radiofrequency (radioastronomy, MRI), microwaves (radar
imaging), infrared wavelengths (thermography), X-rays (medical,
astronomical or industrial imaging) and even gamma rays (nuclear
medicine, astronomical observations).

In addition to electromagnetic imaging, various other modalities are also

employed. These modalities include acoustic imaging (by using infrasound
in geological exploration or ultrasound for echography), electron
microscopy, and synthetic (computer-generated) imaging.

A)Image Storage:

o Mass Storage: Mass storage devices (such as hard drives or

external storage) store the pixel data during image processing.
o Hard Copy Device: Once the image is processed, it can be
stored in a hard copy device (e.g., a pen drive or external ROM
device).
o Purpose: Storage ensures that the processed image is available
for future reference or further analysis.

B) Image Processing:

o Definition: Image processing involves manipulating the pixel

values of an image to enhance its quality, extract information,
or perform specific tasks.
 o Operations: Common operations include contrast
enhancement, noise reduction, edge detection, and feature
extraction.
o Algorithms and Models: Various algorithms and mathematical
models are used to process images effectively.

c) Image Communication:
o Purpose: Image communication involves transmitting or
sharing images across different devices or networks.
o Channels: Images can be communicated via email, social
media, cloud storage, or other communication channels.
o Compression: Image compression techniques reduce the size of
images for efficient transmission.

d) Image Display:

o Monitor or Display Screen: The processed images are

displayed on a monitor or display screen.
o Visualization: Displaying images allows users to visualize the
results of image processing.
o Applications: Image display is crucial in fields like medical
imaging, computer graphics, and multimedia.

20. Quantization

Quantization is a process to convert the continuous analog signal to

the series of discrete values. A quantizer is a device known to perform the
quantization process. The function of quantizer is to represent each level to
the fixed discrete finite set of values.

The signals during transmission over long distances suffer from noise and
interference. To overcome this, the quantization process creates a signal that
is approximately equal to the message signal. It selects a quantized signal
mq(t) with values nearest to the original analog signal m(t). The quantization
process selects a value and rounds off these values to the nearest stabilized
value. The quantized signal mq(t) can get easily separable from the additive
noise.

Let's consider an example of the analog signal confined to the range VA to VB,
as shown below:
21. What is Sampling?

o Definition: Sampling is the process of converting a continuous-

time signal (analog signal) into a discrete-time signal (digital
signal) by selecting specific points at regular intervals.
o Purpose: Sampling allows us to represent continuous signals
using a finite number of samples.
o Sampling Rate (Nyquist Rate): The rate at which samples are
taken is called the sampling rate. It must be at least twice the
highest frequency component in the original signal (Nyquist
theorem) to avoid aliasing.
o Sampling Frequency: The reciprocal of the sampling interval
(time between samples) gives the sampling
frequency (measured in Hz).

22. Define Quantization:

a. Definition: Quantization is the process of mapping
continuous amplitude values (analog levels) to a finite set of
discrete amplitude levels (digital levels).
b. Purpose: Quantization allows us to represent analog signals
using a finite number of digital levels.
c. Quantization Levels: The number of discrete levels
determines the precision of the quantized signal.
d. Uniform Quantization:
i. In uniform quantization, the quantization levels
are equally spaced.
ii. Each step size represents a constant amount of analog
amplitude.
iii. It remains constant throughout the signal.
iv. Examples include A/D converters in digital audio
systems.
 e. Non-uniform Quantization:
i. In non-uniform quantization, the quantization levels
are unequal.
ii. The relation between levels is often logarithmic.
iii. Non-uniform quantization is achieved using techniques
like companding.
iv. Examples include μ-law and A-law quantization used
in telephony.
23. Short note on Comparison:
a. Uniform Quantization:
i. Equally spaced levels.
ii. Simple implementation.
iii. Fixed step size.
b. Non-uniform Quantization:
i. Unequal levels.
ii. Better performance for low-level signals.
iii. Variable step size.

24. What is Image Enhancement?

Image enhancement is the process of making images more useful (such as

making images more visually appealing, bringing out specific features,
removing noise from images and highlighting interesting details in images).

Spatial and Frequency Domains

 Spatial domain techniques manipuletes the pixels of an image

directly. This process happens in the image’s coordinate system,
also known as the spatial domain.

 Frequency domain techniques transforms an image from the

spatial domain to the frequency domain. In this process,
Mathematical transformations (such as the Fourier transform) are
used. The image can be modified by manipulating its frequency
components.
25. What is image segmentation?
Image segmentation is a computer vision technique that partitions a digital
image into discrete groups of pixels—image segments—to inform object
detection and related tasks. By parsing an image’s complex visual data into
specifically shaped segments, image segmentation enables faster, more
advanced image processing.

Image segmentation techniques range from simple, intuitive heuristic

analysis to the cutting edge implementation of deep learning. Conventional
image segmentation algorithms process high-level visual features of each
pixel, like color or brightness, to identify object boundaries and
background regions. Machine learning, leveraging annotated datasets, is
used to train models to accurately classify the specific types of objects and
regions an image contains.

Being a highly versatile and practical method of computer vision, image

segmentation has a wide variety of artificial intelligence use cases, from
aiding diagnosis in medical imaging to automating locomotion for robotics
and self-driving cars to identifying objects of interest in satellite images.
26.what is Simple global thresholding and write use cses?
It is a basic technique used in image segmentation. Let’s explore what it
entails:
1. Definition:
o Simple thresholding is often referred to as global
thresholding because the thresholding function is applied
equally to every pixel of the image, and the threshold value is
fixed.
o The same threshold value is used for each pixel value.
o If the pixel value is less than the threshold value, it is updated
to 0 (black), otherwise, it is updated to the maximum value
(white).
2. Process:
o Given a grayscale image, we choose a threshold value (usually
based on the histogram analysis or domain knowledge).
o Pixels with intensity values greater than the threshold are
treated as white (1) in the output binary image.
o Pixels with intensity values less than or equal to the threshold
are treated as black (0).
3. Use Cases:
o Simple thresholding is effective when the image has consistent
lighting conditions and a clear foreground-background
separation.
o It is commonly used for tasks like object
detection, segmentation, and feature extraction.

27. Distinguish between spatial domain and frequency domain

enhancement techniques.

The spatial domain refers to the image plane itself, and approaches in this
category are based on direct manipulation of pixels in an image. Frequency
domain processing techniques are based on modifying the Fourier
transform of an image. The term spatial domain refers to the aggregate of
pixels composing an image and spatial domain methods are procedures
that operate directly on these pixels. Image processing function in the
spatial domain may he expressed as. g(x, y) = T[f(x, y)] Where f(x, y) is the
input image g(x, y) is the processed image and T is the operator on f defined
over some neighborhood values of (x, y). Frequency domain techniques are
based on convolution theorem. Let g(x, y) be the image formed by the
convolution of an image f(x, y) and linear position invariant operation h(x,
y) i.e., g(x, y) = h(x, y) * f(x, y) Applying convolution theorem G(u, v) = H(u,
v) F(u, v)

28. Explain Derivative Filters and control systems of Derivative action

1. Control Systems - Derivative Action in PID Control:

o In the context of control systems, derivative action is one of the
components of a PID (Proportional-Integral-Derivative)
controller.
o The PID controller aims to regulate a process variable (e.g.,
temperature, pressure, flow rate) to a desired setpoint.
o Here’s a brief overview of the three components:
 Proportional (P) Control: Adjusts the control effort based
on the error (difference between the process variable and
setpoint). It pushes harder when the error is large.
 Integral (I) Control: Compensates for any steady-state
error by integrating the accumulated error over time.
 Derivative (D) Control: Acts as a dampener or brake on
the control effort. It counteracts rapid changes in the
process variable.
o Derivative action helps prevent overshoot and oscillations.
When the process variable approaches the setpoint, the
derivative action slows down the control effort, allowing for
smoother convergence.
o However, applying derivative action requires caution.
Overfiltering (using excessive filtering) can nullify its
benefits. Coordinating the amount of derivative action with the
filtering level is crucial1.
2. Image Processing - Derivative Filters:
o In image processing, derivative filters measure the rate of
change in pixel brightness information within a digital image.
o When applied to an image, derivative filters provide information
about brightness change rates, which can be used for various
purposes:
 Enhancing Contrast: By detecting edges and boundaries,
derivative filters enhance contrast between different
regions in the image.
 Edge Detection: Derivative filters highlight edges (sharp
transitions) between different image features.
  Feature Orientation Measurement: Derivative
information can help determine the orientation of features
(e.g., lines, textures) in the image

29. Image Restoration Using Spatial Filtering



Spatial filtering is the method of filtering out noise from images using a
specific choice of spatial filters. Spatial filtering is defined as the technique
of modifying a digital image by performing an operation on small regions or
subsets of the original image pixels directly. Frequently, we use a mask to
encompass the region of the image where this predefined operation is
performed. A mask is basically a matrix used for computing the new image
pixel intensities from the original one. A mask is used over a small region of
the image and the center pixel value is replaced by the sum of the product
of the original pixel values of the region and the filtering mask coefficients
respectively at those points.
So if we define a 3 x 3 mask as follows:

The mask is specified by its elements and that changes the characteristics
of the mask, resulting in different effects with it. Usually a mask is of odd
order because we have to replace the center pixel intensity value with a new
value obtained by the mask operation.
Filtering creates a new pixel with coordinates equal to the coordinates of the
center of the neighborhood of the image, which the,filter mask covers, and
this center pixel value is equal to the value obtained by the filtering
operation. A filtered output image is generated as the center of the mask
visits each pixel of the image.
30. Smoothing or Averaging Filters
Averaging the pixel intensity values of the image region encompassed by the
mask results in a blurring or smoothing of the image. By replacing the value
of every pixel in an image by the average of the intensity levels enco in the
neighborhood encompassed by the filter mask, the process results in an
image with reduced sharp transitions in intensities, hence an overall
blurring or smoothing effect. This helps in reducing unwanted noise in the
image.
Averaging filter masks can be defined as below.

The above filter mask response can be expressed simply as

Let Sxy represent the set of coordinates in the image region encompassed by
the mask. So the normal averaging filter, also known as the arithmetic mean
filter, can alternatively be represented as :

Another way is to have a weighted average.

The most general weighted average filtering of an M x N image using a mask
of size m x n is given by the following expression:
where x = 0,1,2,…,M-1 and y = 0,1,2,…,N-1; and m = 2a+1 and n = 2b+1.
31. Split and merge segmentation
It is an image processing technique used to segment an image. The image
is successively split into quadrants based on a homogeneity criterion and
similar regions are merged to create the segmented result. The technique
incorporates a quadtree data structure, meaning that there is a parent-
child node relationship. The total region is a parent, and each of the four
splits is a child.
Algorithm

 Define the criterion to be used for homogeneity

 Split the image into equal size regions
 Calculate homogeneity for each region
 If the region is homogeneous, then merge it with neighbors
 The process is repeated until all regions pass the homogeneity test
Homogeneity
After each split, a test is necessary to determine whether each new region
needs further splitting. The criterion for the test is the homogeneity of the
region. There are several ways to define homogeneity, some examples are:

 Uniformity- the region is homogeneous if its gray scale levels are

constant or within a given threshold.
 Local mean vs global mean - if the mean of a region is greater than
the mean of the global image, then the region is homogeneous
 Variance - the gray level variance is defined as

where r and c are row and column, N is the number of pixels in the region

and
An example incorporation would be that the variance of a region be less
than a specified value in order to be considered homogeneous.
32. Region and Edge Based Segmentation



Segmentation

Segmentation is the separation of one or more regions or objects in an

image based on a discontinuity or a similarity criterion. A region in an
image can be defined by its border (edge) or its interior, and the two
representations are equal. There are prominently three methods of
performing segmentation:
 Pixel Based Segmentation
 Region-Based Segmentation
 Edges based segmentation

Edges based segmentation

Edge-based segmentation contains 2 steps:

 Edge Detection: In edge detection, we need to find the pixels that
are edge pixels of an object. There are many object detection
methods such as Sobel operator, Laplace operator, Canny, etc.
 1 0 -1

2 0 -2

1 0 -1

Sobel vertical Operator

+1 2 1

0 0 0

-1 -2 -1

Sobel Horizontal Operator

0 -1 0

-1 4 -1

0 -1 0

Negative Laplace Operator

 Edge Linking: In this step, we try to refine the edge detection by

linking the adjacent edges and combine to form the whole object.
The edge linking can be performed using any of the two methods
below:
 Local Processing: In this method, we used gradient and
direction to link the neighborhood edges. If two edges have
a similar direction vector then they can be linked.
 Global processing: This method can be done using HOG
transformation
 Pros :
 This approach is similar to how the humans brain
approaches the segmentation task.
 Works well in images with good contrast between object
and background.
 Limitations:
  Does not work well on images with smooth transitions and
low contrast.
 Sensitive to noise.
 Robust edge linking is not trivial and easy to perform.

Region-Based Segmentation

In this segmentation, we grow regions by recursively including the

neighboring pixels that are similar and connected to the seed pixel. We use
similarity measures such as differences in gray levels for regions with
homogeneous gray levels. We use connectivity to prevent connecting
different parts of the image.
There are two variants of region-based segmentation:
 Top-down approach
 First, we need to define the predefined seed pixel. Either
we can define all pixels as seed pixels or randomly chosen
pixels. Grow regions until all pixels in the image belongs
to the region.
 Bottom-Up approach
 Select seed only from objects of interest. Grow regions
only if the similarity criterion is fulfilled.
 Similarity Measures:
 Similarity measures can be of different types: For the
grayscale image the similarity measure can be the
different textures and other spatial properties, intensity
difference within a region or the distance b/w mean value
of the region.
 Region merging techniques:
 In the region merging ging technique, we try to combine the
regions that contain the single object and separate it from
the background.. There are many regions merging
techniques such as Watershed algorithm, Split and merge
algorithm, etc.
 Pros:
 Since it performs simple threshold calculation, it is faster
to perform.
 Region-based segmentation works better when the object
and background have high contrast.
 Limitations:
 It did not produce many accurate segmentation results
when there are no significant differences b/w pixel values
of the object and the background.
Implementation:

 In this implementation, we will be performing edge and region-

based segmentation. We will be using scikit image module for that
and an image from its dataset provided.