Scribd
Upload
14 views9 pages

Coursework - Chapter 1

The document discusses land use and land cover classification and change detection using remote sensing. It describes various classification algorithms like ISO cluster, K-means, maximum likelihood classifier, and support vector machines. The objectives are to generate land use maps, assess land cover changes, and compare classification methods like scalar-based and object-based approaches.

Uploaded by

Manish Tiwari
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
14 views9 pages

Coursework - Chapter 1

The document discusses land use and land cover classification and change detection using remote sensing. It describes various classification algorithms like ISO cluster, K-means, maximum likelihood classifier, and support vector machines. The objectives are to generate land use maps, assess land cover changes, and compare classification methods like scalar-based and object-based approaches.

Uploaded by

Manish Tiwari
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 9

Chapter I Introduction

1.1 Introduction

Currently machine learning algorithms are used everywhere for regression and classification such
as supervised and unsupervised classification, we have to test and evaluate new algorithms for
classification other than remote sensing algorithms such as support vector machine from machine
learning for improvement of accuracy,

Image classification is the way toward allotting pixels of consistent raster picture to predefine
arrive cover classes. It is a complex and tedious process, and the after effect of arrangement is
probably going to be influenced by different components (e.g. nature of input images,
classification methods, algorithm, etc.). Keeping in mind the end goal to enhance the
classification exactness, in this manner, determination of suitable arrangement strategy is
required. Pixel and object-based as well as advanced classification approaches. This object-based
approach considers group of pixels and the geometric properties of image objects. It segments the
imageries into homogenous regions based on neighbouring. Thus, an object-oriented method has
been applied to avoid the mixed pixel troubles.

Here we have done study of

1) ISO Cluster
2) K Means
3) Maximum Like hood Classifier
4) Mahalanobis Classifier
5) Minimum Distance Classifier
6) Parallelepiped Classifier
7) Support Vector Machine
1) Radial Basis Function
2) Sigmoid
3) Linear

1
4) Polynomial

1.2 Remote Sensing

Remote detecting implies estimation of some property or getting information of an object


that is not in physical contact with the protest.

1.2.1 Evolution of Remote Sensing

1) 1858: - Aerial photograph using hot air balloon


2) 1887-1888: - Aerial photograph using kite.
3) 1903: - Pigeon mounted camera
4) 1908-1945: - Aero plane were used
5) 1947: - First space-based photo
6) 1972: - Landsat 1 First Satellite
7) 1980: - Hyper spectral, Microwave, Radar

1.2.2 GIS & Remote Sensing


In past when there were no computers and remote sensing the LULC change was detected
using tracing paper and topographic sheet. But this method was very time consuming and tedious
for larger area and took big effort. So, this method is suitable for accessing such highly critical
areas, and monitors their periodic changes. To understand ground reality, we have to prepare the
maps from statistics generated by algorithms needs GIS software’s. [5] D

SUN

A
2

G
Figure 1.1: Process of remote sensing

A) Source of Energy: - Wellspring of vitality which lights up or gives electromagnetic


vitality to the objective of intrigue.
B) Atmosphere: - Vitality heads out from its source to the objective; it will interact with and
communicate with the environment it goes through.
C) Interaction with the Target: - Once the vitality advances toward the objective through
the climate, it cooperates with the objective relying upon the properties of both the
objective and the radiation.
D) Recording of Energy by the Sensor: - After the vitality has been scattered by, or
discharged from the objective, we require a sensor to gather and record the
electromagnetic radiation.
E) Transmission, Reception, and Processing: The vitality recorded by the sensor must be
transmitted, regularly in electronic shape, to an accepting and handling station where the
information are prepared into an image.
F) Interpretation and Analysis: - The processed image is interpreted, visually and/or
digitally or electronically, to pull out information about the object.
G) Application - The last component of the remote sensing uncovers some new data, or help
with tackling a specific issue. [3] [5]
1.3 Change Detection
Land use land cover change is critical issue now a day because the fact that there is quick
development and extension of city, shortage of land, at that point need of more creations LULCC
react to financial, political, social, statistic and ecological conditions and powers which are to a
great extent described by high human populaces. LULCC has turned out to be one of the
significant worries of analysts and chiefs around the globe today. Here we have demonstrated
change location of following essential factors, for example,

3
1) Agricultural Land 5) Settlement

2) Water Body 6) Barren Land

3) Vegetation

4) Follow Land

1.5 Electromagnetic Spectrum

Co-operation of electromagnetic radiation with surfaces makes EMR be reflected or


transmitted into the surface and sub-surface and all incompletely, re-emanated or scattered. Every
one of the procedures is firmly reliant on the wavelength of the occurrence radiation and also the
nuclear and sub-atomic structures of the material sand its condition. The measure of vitality
reflected back to a sensor relies upon the wavelength and power of the episode vitality, the
surface unpleasantness, the introduction of the objective and the objective's ingestion qualities as
for wavelength reflection; transmission, assimilation and discharge are frequently regarded as
surface marvels. Everything in the nature has its own supply of reflected, transmitted and
assimilated radiation.

Figure 1.2: Electromagnetic spectrum


1.6 Resolution

4
The minimum distance between two objects, means resolving power to identify not only
presence of two objects but also their properties. There are different resolutions listed below.

1) Spatial
2) Spectral
3) Hyper spectral
4) Radiometric
5) Temporal
1.6.1 Spatial Resolution

It is measure of area or size; it is defined by the size of the pixel on the ground in meters. e.g. 30
M, 15M, 5M, 1M

1) Very high resolution: - 1M

2) High resolution: - 5M

3) Medium resolution: - 15M

4) Low resolution: - 30M or 69M

1.6.2 Spectral Resolution

A multispectral image made up of no of bands. For visualization, grayscale image comes with
one band and combining of different bands at a time as a create colour composite image.
Following Figure shows that three visible bands in multispectral resolution.

Figure 1.3: Multispectral resolution


1.6.3 Hyper Spectral Resolution

A hyper spectral image consists of more no of bands as compared to spectral resolution.


Following Figure shows that three visible bands in hyper spectral resolution.

5
Figure 1.4: Hyper spectral resolution

1.6.4 Radiometric Resolution

Measure of how many grey levels are measured between pure black. It is expressed in
bits. e.g. 28 means 256 DN values starting from 0 to 255 DN values are present. More no of bits
fine the image and less no of bits coarse image. The below table describe details about bits and
DN values.

Table 1.4: Bits and DN values


No of Bits 1 2 3 4 5 6 7 8 9 10 11 12
Maximum DN Values 2 4 8 16 32 64 128 256 512 1024 2048 4096

The below table tells us about different sensors and their properties such as swath sensor
and radiometric resolution

Table 1.5: Sensor and Resolution


Mission Sensor Radiometric Swath
Resolution
Resourcesat2 LISS3 10 140KM

ETM + ETM + 8 Bits 185KM

Landsat 8 OLI 12 Bits 185KM

Landsat 8 TIRS 12 Bits 185KM

1.6.5 Temporal Resolution


6
Transient determination is characterized as the measure of time expected to return to and gain
information for precisely the same.

1.7 Aim & Objectives of Study


The aim of this research is land use land cover classification and change analysis by multi
temporal remote sensing data. The aim will be supported by following objectives. The main
objectives are given below:

1) Generating LU/LC maps and statistical data of land consumption of Nanded.


2) To assess implications of changes observed in study area and make appropriate
recommendations.
3) Provide platform for environmental studies alarmed with LU/LC Change.
4) Study of unsupervised classification with LU/LC mapping.
5) Study of best scalar based supervised classification with accuracy assessment.
1. Maximum likelihood classifier
2. Mahalanobis classifier
3. Minimum distance classifier
4. Parallelepiped classifier

6) Study of best object-based classifier with accuracy assessment


1. Support Vector Machine
1.1 Radial basis function
1.2 Sigmoid
1.3 Linear
1.4 Polynomial
7) Comparative analysis of scalar based and object-based classification.

7
1.8 Motivation
Currently machine learning algorithms are used everywhere for regression and
classification such as supervised and unsupervised classification, we have to test and evaluate
new algorithms for classification other than remote sensing algorithms such as support vector
machine from machine learning for improvement of accuracy, so we used support vector
machine for classification where we have achieve better accuracy and R Programming Language
for statistical analysis and chart preparation.

1.9 List of Abbreviations


ETM Enhanced Thematic Mapper
GCP Ground Control Points
PAN Panchromatic
GIS Geographic Information System
RS Remote Sensing
LISS 3 Linear Imaging Self Scanner
LU Land Use
LC Land Cover
GCP Ground Control Points
LULCC Land Use Land Cover Change

8
9

You might also like