Escolar Documentos
Profissional Documentos
Cultura Documentos
A Class Project
By
Mayuri Bhattacharyya
Advisor
The region is in the central east part of the state of Chihuahua in Mexico. This is a watershed
region bounded by two highlands of Mexico namely Sierra Bonce in the north and Sierra la
Gloria in the south. The name of the river is Sierra del Soldado which is one of the major streams
in Mexico.
This river is important in respect to the water distribution system with in Chihuahua in Mexico.
This river is the perennial, and meandering in its course. The above three maps are of Mexico,
Chihuahua State, and the study area, respectively. My current research is on this region’s images
which include a remote sensing application for image classification of the study area using
unsupervised methods.
The above images are the satellite imagery of the study area using TM (Thematic Map) bands.
The first image is a combination of three part of the same image. The first bigger part (Image
window) shows the main study area. The left side image with the black boundary (Scroll
window) shows the surrounding area in a broad way, and the right side small image (Zoom
window) shows the particular part of the image in a zooming layer. These images are the product
of the remote sensing software ESRI ENVI 4.2. The left side broad image is from the Google
earth software.
Hypothesis:
I have attempted to make a remote sensing application for image classification for the study area.
Why this effort is required because this region is neglected for the study of scientific research
due to its inaccessibility for the steep surface with the existence of
some dense forest. Therefore remote sensing applications should
be applied to study this region, to extract the thematic information
from the images, and to recognize the pattern of the bands in the
images. My research is going to prove if we follow the remote
sensing techniques we must be careful about the selection of the
classification methods because there might be some
misinterpretations, too. I am going to show use of unsupervised
approaches to classify the images of the study area. I have used TM images with three bands,
namely: Red (R) band 4, Green (G) band 3, and Blue (B) band 2.
Methods:
Now a day people generally rely on the multispectral classifications, algorithms based on
parametric and nonparametric statistics using either supervised or unsupervised logic. Parametric
methods such as maximum likelihood classification and unsupervised clustering assume
normally distributed remoter sensor data and knowledge about the forms of the underlying class
density functions. I have used here unsupervised classification methods to explore the spectral
characteristics.
Unsupervised Classification-
The clustering process results in a classification map consisting of m spectral classes. The analyst
then attempts a posteriori (after the fact) to assign or transform the spectral classes into thematic
information classes of interest like forest, agriculture, etc. This may be difficult. Some spectral
clusters may be meaningless because they represent mixed classes of Earth surface materials.
The analyst must understand the spectral characteristics of the terrain well enough to be able to
label certain clusters as specific information classes. Unsupervised classification can be used to
cluster pixels in a data set based on statistics only, generally two techniques are in common use:
K- Means and
ISODATA
Applications:
The remote sensing software ENVI has integrated spectral profiling capability to examine the
spectral characteristics of the data which is very important for
analyzing the association of the cluster of the pixels in the
image. For this understanding I have used the Z profile tool
(Spectrum) to extract the spectral profile of the image. The
graph is known as Spectral Plot which shows the spectral
profile (Z profile) of the image.
The second one is showing the spectrum of the Luis L. Leon Lake where the association of blue
color is darkest and maximum, and the other bands are almost absent.
The third plot is on the river meander of Sierra del Soldado where maximum red band is
accumulated.
The fourth plot is near the region of lake where maximum white color is viewed this is because
the mixture of RGB forms white. So in that part of the region shows a combination of all three
bands. Following are the images and the Spectral plots-
(a) Sierra Bonce – The Highland:
(b) Luis L. Leon (El Granero) – The Lake:
(c) Sierra del Soldado – The River Meander:
The K-Means unsupervised classification calculates initial class means evenly distributed in the
data space then iteratively clusters the pixels into the nearest class using a minimum distance
technique. Each iteration recalculates class means and reclassifies pixels with respect to the new
means. All pixels are classified to the nearest class unless a standard deviation or distance
threshold is specified, in which case some pixels may be unclassified if they do not meet the
selected criteria. This process continues until the number of pixels in each class changes by less
than the selected pixel change threshold or the maximum number of iterations is reached.
“File: D:\mb\TM
images\TM2003\LE7032040000311550.H1
Dims: 7477 x 7075 x 6 [BSQ]
Size: [Byte] 317,398,650 bytes.
File Type: NLAPS CD
Sensor Type: Landsat TM
Byte Order: Host (Intel)
Projection: UTM, Zone 13 North
Pixel : 30 Meters
Datum : WGS-84
Wavelength: 0.485 to 2.215
Upper Left Corner: 1, 1
Description: Landsat TM NLAPS File
Imported into ENVI [Thu Dec 14
07:28:25 2006]
Total Cache Size: 25,165,824 bytes
In-memory Use: 0 bytes
Available : 25,165,824 bytes
Output Size : 52,899,775 bytes
This result outputted to memory will EXCEED the total cache size. Are you sure you want to
output this item to memory?”
Following is the image of the study area classified by the ISODATA Technique.
Original Image
Screen Data
Location Display ID R G B R G B
28°53'8.03"N,
105°18'13.40"W 59,493,466 122 136 137 86 150 20
28°52'24.91"N,
105°25'35.15"W 55,503,509 114 0 0 84 57 66
28°56'27.64"N,
105°25'36.14"W 55,503,260 61 88 90 70 124 102
28°50'4.54"N,
105°25'34.58"W 55,503,653 95 121 126 79 142 116
28°50'4.76"N,
105°18'12.86"W 59,493,654 80 118 116 75 140 112
28°56'33.71"N,
105°18'15.10"W 59,483,255 26 56 69 61 106 94
Cursor Location Data values for specific locations on the Original Image:
Cursor Location Data values for specific locations on the Image after K-Means Classification:
ISODATA
Screen Data
Location Display ID R G B R G B
28°53'8.03"N,
105°18'13.40"W 59,493,466 0 255 255 - - -
28°52'24.91"N,
105°25'35.15"W 55,503,509 0 0 255 - - -
28°56'27.64"N,
105°25'36.14"W 55,503,260 0 255 255 - - -
28°50'4.54"N,
105°25'34.58"W 55,503,653 255 0 255 - - -
28°50'4.76"N,
105°18'12.86"W 59,493,654 255 0 255 - - -
28°56'33.71"N,
105°18'15.10"W 59,483,255 255 0 0 - - -
Cursor Location Data values for specific locations on the Image after ISODATA Classification: