Satellite Image Processing

and Analysis

O
R
DR. B. KRISHNA MOHAN

/IS
CENTRE OF STUDIES IN RESOURCES ENGG
INDIAN INSTITUTE OF TECHNOLOGY BOMBAY
[email protected]
sc
nr
 About CSRE

 Centre of Studies in Resources Engineering at IIT Bombay

offers M.Tech. and PhD programmes in the area of

O
Geoinformatics and Natural Resources Engineering

R
 Started as a Centre of Excellence in 1976 to conduct

/IS
research and development in remote
sensing, GIS, satellite image processing and applications
sc
to natural resources exploration and management
nr
 Became a full fledged teaching Centre in 2005
 CSRE Faculty Specializations

 Faculty members drawn from multitude of

disciplines –

O
 Mathematics

R
 Physics

/IS
 Electrical Engg.

 Computer Science
sc
 Civil Engg.

 Earth Sciences
nr

 Agriculture and Soil Sciences

 Analytical Chemistry
 CSRE Research Areas
 Theory and Algorithms
 Optical, Microwave and Hyperspectral Remote Sensing
 GPS, GI Systems and GI Science
 Satellite Image Analysis, Stereo vision

O
 Applications

R
 Terrain Evaluation and Landuse Planning, Drought

/IS
 Urban Planning, Environment and Applications
 Surface and Ground Water
sc
 Snow/Ice mapping
nr
 Rural Development
 Agro-informatics and Precision Farming
 Avalanche and Landslide studies
 Mineral Systems Studies
 O
IRS-1C
LISS3

R
23.5m

/IS
sc
nr
IRSP6 LISS4 Image
O
R
/IS
sc
nr
Quickbird Window

nr
sc
/IS
R
O
 O
AIRBORNE
NATURAL

R
COLOR

/IS
COMPOSITE
sc
nr
Selected Problems in Multispectral Image
Analysis

 Image Segmentation
 Texture Analysis – fractal/Radon transform based

O
 Edge /line Detection – morphological/neural

R
 Image Classification

/IS
 Neural Network Classification
 Genetic Algorithm Pre-processing / Contextual Post-
sc
processing
nr
 Fuzzy Integration for ensemble classifier Combination
 O
R
/IS
sc
IRS1B Image
nr

Genetic-Neural-Classification
 Fractal Textural Classification

O
R
/IS
sc
nr
IRS1C PAN Image
 O
R
/IS
sc
nr

Linear features mixed with building

outlines
Overlay
of
Linear feature
network

O
on

R
classified
image

/IS
sc
nr
 Basic Pattern Radon Spectrum

O
Radon Principle

R
/IS
sc
nr

Sand dunes – IRS1B Radon Texture

 Selected Problems

 Image Compression
 Rate constrained JPEG for custom rate of compression

O
 Change Detection

R
 Super-pixel approach

/IS
 Texture based approach
sc
nr
 Quantization Tables

1 1 1 1 1 2 3 5
1 1 1 1 2 3 5 5
1 1 2 2 3 5 5 7
1 2 2 3 5 5 6 7
2 2 3 4 6 6 7 23

O
2 3 4 6 6 7 20 32
3 4 6 6 9 16 38 68

R
Uncompressed MUMBAI(11 BIT) CR = 3.5 4 6 6 12 15 65 67 71

/IS
2 8 9 13 13 21 22 41

8 10 12 13 21 22 40 43

10 12 15 20 23 40 44 55
sc
11 15 20 23 39 46 54 55

17 17 25 35 46 54 56 83
nr
17 25 35 49 51 57 63 88

28 32 50 51 59 61 104 115

30 50 50 59 61 105 112 127

CR = 6 CR = 9.5
 High Spatial Resolution Image Analysis

 Development of a Generic Framework

 Complete System to address requirements of high spatial

O
resolution image analysis

R
 Development of tools

/IS
 Segmentation

 Component Labeling
sc
 Feature Computation

 Classification
nr
 Pre-process

Decompose image
Generic High
Spatial Resolution Segment image at Different Single stage
Resolutions
Image Analysis segmentation

O
Framework

R
Link Segments

/IS
Connected Comp. Labeling
sc
nr
Spatial Features Spectral Features Texture Features Context

Object Specific Classification General purpose LU/LC

classification
 Building Extraction

O
R
/IS Road Extraction
sc
nr

General Image Segmentation

and Classification
 Hyperspectral Image Analysis

 Development of a Generic Framework

 System to address requirements of high spatial resolution

O
image analysis

R
 Development of selected tools

/IS
 Dimensionality Reduction

 Classification
sc
 Ongoing work
 Search Techniques for Feature Selection
nr

 Super-resolution
 O
R
/IS
sc
nr
Hyperspectral Image Window
and Spectrum of Vegetation
near Powai Lake
 High spectral resolution image
High Spectral
Resolution Image Atmospheric Correction

Analysis

O
Dimensionality Reduction

R
/IS
Pure Pixel / Training Data
Spectral
Identification
sc libraries

Supervised Mixture Modeling Spectral

nr
Classification Matching

Abundance Mapping
General Purpose
Classification
classification
 Kernel Classification

O
R
Class 1 Black – Water
Class 2 Blue – Barren land

/IS
Class 3 Green - Vegetation
Class 4 Red – Moist land
sc
Class 5 Yellow – Vegetation 2
nr

23
 Educational Content Development
 Multimedia Image Processing Tutor
Content

 Demonstrations

O
 Resources
 Dedicated Image Processing Package

R
 Virtual Lab

/IS
 Collection of software experiments
 Cover theory, procedure, execution and self-assessment
sc
 Optical/infrared part ready
 Work in progress on SAR
nr

 Work on Hyperspectral Image Analysis at planning stage

 http://vlabs.ac.in/siplabs (also,
 http://virtual-labs.ac.in/siplabs)

nr
sc
/IS
R
O
 Tutor

DIGITAL IMAGE PROCESSING FOR REMOTE SENSING

Subject Animated Hands-on Additional Glossary and

Animated
Content Demos Software Resources Exercises
Demo

O
R
Basic concepts FCC Links to online RS and IP

/IS
of DIP for RS resources terminology
Enhancement ReSIPro
Advanced Software List of Hands_on with
Filtering books, journals
concepts of
sc ReSIPro for IP
DIP for RS Classification and IP exercises
products
(35 chapters) Compression
nr
Case studies

Key features: 1. Relevant to typical university syllabi; 2. More than an awareness CD;
3. Useful as a detailed self-study material; 4. Self-contained with
software, demonstrations, case studies, sample data, references, links to online
resources, hands-on exercises and self-assessment quizzes and suggested student
26
projects
 Educational Outreach

 Introductory Short Term Courses for College

Teachers

O
 Basic Course to expose potential of remotely sensed images

R
 Introductory Satellite Image Processing

/IS
 Hands-on Sessions
 Advanced Courses for College Teachers
sc
 Modern Tools like Neural Networks, Support Vector Machines

 Image Segmentation Tools

nr

 Multiresolution Image Analysis

 Image Registration and Change Detection

 Educational Outreach

 Invited Lectures in different colleges/universities

 Introduction to satellite image processing

O
 Texture analysis methods

R
 Image classification techniques

/IS
 Image compression for spaceborne applications

 Neural Networks, Support Vector Machines

sc
 Image Segmentation

 Hyperspectral image analysis

nr
GIS Related Research

O
R
INDIGENOUS TECHNOLOGY

/IS
APPLICATION RESEARCH
sc
nr
GRAM++ Desktop GIS

O
R
/IS
sc
nr

www.bhugolgis.com
 GIS Related Research

 Geospatial Data Security

O
 Vector data watermarking

R
/IS
 Raster data watermarking
sc
nr
 Challenges

 Vector data watermarking

O
 Strategies for point, line and polygon layers

R
 Determining capacity of a layer for watermarking

/IS
 Single and multiple watermarking
sc
nr
 Challenges

 Raster data watermarking

O
 Efficient data structures for handling large sized

R
spatial data

/IS
 Single and multiple watermarks
 Strategy for multiple watermarking
sc
nr
 Summary

 CSRE is a vibrant academic unit of IIT Bombay

 Research and development in the multidisciplinary

O
area of geospatial technologies

R
 Extensive user of IRS series data of all types

/IS
 Applications to natural resources, environment,
engineering problems, disaster management
sc
 Capacity building
nr

 Active industry interaction through consultancy

 Bright students for M.Tech. and PhD are welcome!
 O
R
THANK YOU!
/IS
sc
nr