Professional Documents
Culture Documents
REMOTE SENSING
Professor Sabyasachi Maiti
Department of Geology & Geophysics
I I T Kharagpur
SUBMITTED BY
YASH LADHA
10GG20038
Final Year Student,
Department of Geology and Geophysics
Procedure
From the given satellite image, different features were identified using different land condition.
In the natural land:
1. Water
In 3,2,1 composite(true colour), water appears blue. Water has maximum reflectance in blue land,
i.e. Land 1. In false colour composite (6, 5, 4) water appears as black. The reason is since band 4 is
assigned to channel 1 instead of band 1, water absorbs infra-red radiation thus appearing black.
The different water bodies are: a) Ocean, b) Meandering river, c) Oxbow lake
2. Vegetation
In the land combination (4, 3, 2), vegetation appear red. The greatest amount of electromagnetic
energy reflected by plants is near IR
4-IR
Shown by Red colour
3-Red
Shown by green colour
2-Green
Shown by blue colour
Natural
Oxbow Lake
Mangrove Forest
Chenier Plain
Beach Sand
Meandering point bar
Ocean
Sparse Vegetation
Meandering river
Yash Ladha (10GG20038)
GEO-REFERENCING
It is a process to relate a map with real world co-ordinates. This establishes the location in terms of coordinate system.
Procedure
1. Open the satellite image in ERDAS IMAGINE, (A)
2. Open the topo-sheet which is already geo-referenced.
3. In the 1st viewer (A),
Click as follows
Raster Geometric Correction Polynomial. Click on Add/Change Projection
Projection Type
Spheroid Name
Datum Name
Choose polynomial model
UTM
WGS-84
Zone 45
The Ground Control Point (GCP) tool box will soon open.
5 GCPs were chosen and equally distributed
Result
The new file was geo-referenced based on the topo-sheet.
IMAGE CLASSIFICATION:
These processes group pixels in a digital image to represent land cover features. It is the labelling of image in
pre-defined categories.
Supervised:
The user needs to select training sites from land cover. This classification is based on spectral signature.
Unsupervised:
Pixels are grouped on the reflectance properties of pixels. These groupings are called clusters. The user
needs to identify the number of clusters to generate and which land to use.
Process for Supervised Classification
1. Open the image.
2. Click on data preparation Supervised classification.
IMAGE ENHANCEMENT:
This is a process via which visual interpretation of image becomes easy. Enhancement makes more of the
range useful, since the useful data lies in the range (0-255).
Method
1. Contrast Enhancement
At first image is opened in the viewer in a single band (i.e. Greyscale)
Click as follows
Raster Contrast Breakpoint Points are pressed on lower Apply and upper end of
histogram.
The image histogram spreads to 0-255 and the enhanced image is seen in a viewer.
2. Histogram Equalization
Click as follows
Interpreter Histogram Equalization Select input file Choose a name for output OK
Then displays the image in a viewer and compare with linearly stretched image and its histogram.
Histogram stretching is done based on cumulative histogram values. Highly populated pixels are
stretched more compared to less populated pixels.
3. Filtering :
At first image is opened in a viewer.
Click as follows
Raster Filtering Low Pass Smoother, High Pass Sharpen (Edge Enhancement),
Convolution
Each option is selected.
The changes are seen in image.
BAND COMBINATION
In a viewer, click as follows