Sei sulla pagina 1di 30

-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?

Potrebbero piacerti anche