RAJIV GANDHI PROUDYOGIKI VISHWAVIDYALAYA,

BHOPAL New Scheme Based On AICTE

Flexible Curricula Computer Science and Engineering,

Scan for YouTube

UNIT ll

What are the Image Representation and Description:

In computer vision, image representation and description refer to the process of extracting
features and characteristics from an image and converting them into a format that a computer
can process and understand. There are different representation schemes and boundary
descriptors used in image processing.

Representation schemes refer to the methods used to extract features from an image, which are
then used to describe or identify the im
image.
age. Some of the most common representation schemes
include:

1. Pixel-based
based representations: This involves using the raw pixel values of an image as its feature
vector. This is a simple but effective way to represent an image, although it can be sensitive to
changes in illumination and image noise.
2. Frequency-based
based representations: This involves converting an image into the frequency domain
using techniques like Fourier transforms. The frequency domain representation can be used to
identify patterns or textures in an image.
3. Texture-based
based representations: This involves using statistical methods to extract texture features
from an image. Texture features describe the variation in intensity or color in different regions of
an image.
4. Shape-based
based representations: Thi
Thiss involves using geometric methods to extract shape features
from an image. Shape features describe the boundaries and contours of objects in an image.

Boundary descriptors are a type of shape

shape-based
based representation that describe the edges and
boundaries of objects in an image. There are several types of boundary descriptors, including:

1. Chain codes: This involves encoding the boundary of an object as a sequence of directions that
the boundary takes at each pixel.
2. Fourier descriptors: This involves representin
representingg the boundary of an object as a sum of cosine and
sine waves.
3. Curvature scale space: This involves representing the boundary of an object as a series of curves
that describe the shape of the boundary at different scales.

Boundary descriptors are commonly used in tasks such as object recognition, image retrieval,
and image segmentation.

YouTube Channel: RGPV

PV Exam official 2023 https://www.youtube.com/@RGPVExamOfficial
What is Region descriptors Binary Machine Vision

In binary machine vision, image segmentation refers to the process of dividing an image into
different regions or segments based on its features or characteristics. Region descriptors are
methods used to describe the characteristics of these segmented regions. Some common region
descriptors used in binary machine vision include

Thresholding:

Thresholding is a simple and commonly used technique for image segmentation in binary
machine vision. It involves converting a grayscale or color image into a binary image by
selecting a threshold value that separates the foreground pixels from the background pixels. The
threshold value can be selected manually or automatically based on the image histogram.

There are several types of thresholding techniques, including global thresholding, adaptive
thresholding, and Otsu's method. Global thresholding involves using a fixed threshold value for
the entire image, while adaptive thresholding adjusts the threshold value for each pixel based
on the local image intensity. Otsu's method is an automatic thresholding technique that selects
the threshold value that maximizes the separation between the foreground and background
pixels.

Segmentation:

Segmentation is the process of dividing an image into multiple regions or segments, each of
which corresponds to a specific object or part of an object. Segmentation is a crucial step in
image analysis and is used for tasks such as object recognition, image enhancement, and image
compression.

There are several techniques used in image segmentation, including thresholding, edge-based
segmentation, region-based segmentation, and clustering-based segmentation. Thresholding
involves converting a grayscale or color image into a binary image based on a selected
threshold value. Edge-based segmentation involves detecting edges in an image and using
them to define the boundaries of objects. Region-based segmentation involves dividing an
image into regions based on pixel intensity, texture, or other image features. Clustering-based
segmentation involves grouping pixels into clusters based on their similarity in color or intensity.

Connected Component Labeling:

Connected component labeling is a technique used to identify and label individual objects or
regions in a binary image. It involves finding all connected pixels in the image that belong to the
same object and assigning a unique label to each object. Connected component labeling is

YouTube Channel: RGPV Exam official 2023 https://www.youtube.com/@RGPVExamOfficial

 commonly used in image processing tasks such as object recognition, feature extraction, and
object tracking.

There are several algorithms used for connected component labeling, including the two-pass
algorithm, the one-pass algorithm, and the recursive algorithm. The two-pass algorithm involves
two passes through the image to identify and label the connected components. The one-pass
algorithm is a more efficient algorithm that only requires one pass through the image. The
recursive algorithm is a recursive implementation of connected component labeling that uses a
stack or recursion to identify and label connected components.

In summary, region descriptors in binary machine vision involve various techniques such as
thresholding, segmentation, and connected component labeling, which are used for tasks such
as image analysis, object recognition, and feature extraction. These techniques are important
tools for a range of applications in fields such as computer vision, robotics, and medical
imaging.

What are Motion-based segmentation. Area Extraction ?

Motion-based segmentation:

Motion-based segmentation is the process of separating moving objects from the background
in video sequences. It involves detecting and tracking the motion of objects over time and using
this information to segment them from the background. There are several techniques used in
motion-based segmentation:

1. Optical Flow: Optical flow is a technique that estimates the motion of objects by analyzing the
movement of pixels between consecutive frames. It can be used to detect the motion of small
objects and is often used in real-time applications. However, optical flow can be sensitive to
noise and can result in inaccurate motion estimates.
2. Background Subtraction: Background subtraction is a technique that separates moving objects
from the background by subtracting a reference frame from the current frame. It can be used to
detect larger objects and is often used in surveillance systems. However, background
subtraction can be affected by changes in illumination and can result in false detections due to
shadows or reflections.
3. Region-based Methods: Region-based methods involve dividing the image into regions and
analyzing the motion of each region. This method can be used to detect complex objects and
can be more robust to changes in illumination. However, region-based methods can be
computationally expensive and may require manual initialization.

Once the motion-based segmentation has been performed, the segmented regions can be
analyzed and measured.

YouTube Channel: RGPV Exam official 2023 https://www.youtube.com/@RGPVExamOfficial

 Area Extraction:

Area extraction is the process of extracting and measuring the areas of segmented regions
obtained through segmentation techniques. Here are some common concepts, data structures,
and algorithms used in area extraction:

1. Edge Detection: Edge detection involves detecting the edges of objects in an image, which can
be used to identify the boundaries of the segmented regions.
2. Line-Linking: Line-Linking involves linking together edge segments to form complete lines or
contours, which can be used to represent the boundaries of the segmented regions more
accurately.
3. Hough Transform: The Hough transform is a technique used to detect lines or curves in an
image by transforming the image space into a parameter space. It can be used to detect the
boundaries of the segmented regions and to extract features such as the length and orientation
of lines.
4. Line Fitting: Line fitting involves fitting a line to a set of edge segments, which can be used to
represent the boundaries of the segmented regions more accurately.
5. Curve Fitting (Least-Square Fitting): Curve fitting involves fitting a curve to a set of data points,
which can be used to represent the boundaries of the segmented regions more accurately.
Least-square fitting is a common technique used for curve fitting, which involves minimizing the
sum of the squared errors between the curve and the data points.

Once the segmented regions have been extracted, their areas can be measured using various
methods such as pixel counting, contour integration, and moment-based methods. These area
measurements can be used for tasks such as object tracking, object recognition, and motion
analysis in video sequences.

YouTube Channel: RGPV Exam official 2023 https://www.youtube.com/@RGPVExamOfficial