-Richa Sinha
100280723003
IHS based Pan Sharpening with
Semester 3rd
Java Image Processing
M. E. (Information Technology)
Agenda
Geo-Image Processing
Pan Sharpening
Imaging done through JAI.jar
The Algorithm
An Example on Java
Reference
Agenda
Geo-Image Processing
Pan Sharpening
Imaging done through JAI.jar
The Algorithm
An Example on Java
Reference
Introduction
Processing done on Spatial data
In satellite imaging two images are produced
First one uses various spectral bands to create
false or true images of lower resolution
Second one is panchromatic images of higher
resolution
Types of Geo-Image Processing
Geometric correction
Image matching
Image calibration
Contrast enhancements
Hardcopy and digital output in client specific
formats
Agenda
Geo-Image Processing
Pan Sharpening
Imaging done through JAI.jar
The Algorithm
An Example on Java
Reference
Image types
RGB Color Image format
Combination of 3 band i.e.
Red Green Blue format
Panchromatic Image
Single band Image
Other Color Image format
CMY format
IHS format
Many More.
Image fusion
Combining two images to get a single image
Main advantage : To convey more information
precise and clear information
Types
High Resolution Imagery
Pan Sharpening with False Colour Image
High Resolution Imagery
Pan Sharpening with False Color Image
Agenda
Geo-Image Processing
Pan Sharpening
Imaging done through JAI.jar
The Algorithm
An Example on Java
Reference
Imaging done with JAI
Obtain the image
Define the imaging graph
Evaluate the graph
Process the result
Agenda
Geo-Image Processing
Pan Sharpening
Imaging done through JAI.jar
The Algorithm
An Example on Java
Reference
Step 1: Select the image bands
used to create a RGB image
3 different spectral bands combine to form an
RGB image
Red Band, Green Band and Blue band merge
to form white color image which has 3 bands
of RGB format
Step 1 Continue
3 bands pixel values are
taken
A matrix is taken which is
(# of source bands + 1) /
(# of desired destination
bands)
Dst = (255 * 0.25) + (157
* 0.5) + (28 * 0.75)
Bandmerge does the job
in JAI
Step 2 : Rescale the RGB image
equal to panchromatic
RGB images image
size = Panchromatic image size
Scaling factors
ScaleX = width(rgbimage) / width(panimage)
ScaleY = height(rgbimage) / height(panimage)
ScaleX and ScaleY passed as parameter
Scaling is done by
x(dstwidth) = x(srcwidth) * scaleX
y(dstheight) = y(srcheight) * scaleY
scale does this job of scaling in JAI
Step 3 : Convert the RGB image to
IHS based
IHS Intensity -color
Hue - model
Saturation Model
RGB to IHS conversion
I = (R+G+B)/3
S = 1-min(R, G, B)/I
H = G > B ? h : 2*PI - h, where
h = cos-1{[(R-G)+(R-B)]/2[(R-
G)2+ (R-G)(G-B)]1/2}
Colorconvert is the operator
which helps in this conversion
inside IHSColorModel Class
Step 4 : Split the IHS image into its
three components
IHS image is split into its three bands
Intensity band(band 0)
Hue band(band 1)
Saturation(band 2)
BandIndices parameter
The destination pixel values will be
dst[x][y][b] = src[x][y][bandIndices[b]];
bandselect does this job JAI
Step 5 : Substitute I band in the
image by the panchromatic image
Intensity band (band 0 ) is replaced with the one
band panchromatic image
Advantage:
Panchromatic Image is correct resolution image so
by replacing Intensity band with this band, the
correct resolution image
Replacing Hue or Saturation band will not correct
picture because they are related to color
Step 6 : Merge the three bands to
a single
The 3 bandsIHS image
which will be combined again are
Panimage band
Hue band
Saturation band
The band merging will take place again in same as
RGB bands were combined
Again bandmerge operator is called for JAI
Step
Reverse7 : Convert
Conversion of IHS model tothe IHS to the
RGB model
When H is in [2PI/3, 4PI/3], R should be the smallest. Then,
RGB color model
R = (1-S)I and
G = (c1 + c2)/2
B = (c1 - c2)/2 where c1 = 3I-R and c2 = sqrt(3)(R-I)tg(H)
When H is in [4PI/3, 2PI],
G = (1-S)I and
B = (c1 + c2)/2
R = (c1 - c2)/2 where c1 = 3I-G and c2 = sqrt(3)(G-I)tg(H-2PI/3)
When H is in [0, 2PI/3],
B = (1-S)I and
R = (c1 + c2)/2
G = (c1 - c2)/2 where c1 = 3I-B and c2 = sqrt(3)(B-I)tg(H-4PI/3)
Agenda
Geo-Image Processing
Pan Sharpening
Imaging done through JAI.jar
The Algorithm
An Example on Java
Reference
An Example
Agenda
Geo-Image Processing
Pan Sharpening
Imaging done through JAI.jar
The Algorithm
An Example on Java
Reference
References
http://java.sun.com/products/java-media/jai/f
orDevelopers/jai1_0_1guide-unc/Preface.doc.ht
ml
http://www.pcigeomatics.com/
http://www.geosage.com/
http://www.wikipedia.org/
Questions?
