Artificial Intelligence For Mixed Pixel Resolution By Nitish Gupta (Guru 1
Feb 03, 2016
Artificial Intelligence For Mixed Pixel Resolution
By
Nitish Gupta (Guru Gobind Singh Indraprastha University)
Dr. V.K. Panchal (Defence Research Development Organization)
1
OutlineOutline
27-JULY-2011IGARSS,2011-VANCOUVER
2
3IGARSS,2011-VANCOUVER
Conflicts are one of the most characteristic attributes in Satellite Remote Sensing multilayer imagery.
Class conflict occurs when there is presence of spectrally indiscernible distinct classes and how the human experts understand it based on his/her expertise.
Can we resolve those mixed pixels ?
\
27-JULY-2011
SPATIAL RESOLUTION & MIXED PIXEL
100Meter resolutionPatalganga, India
4
27-JULY-2011IGARSS,2011-VANCOUVER
SPATIAL RESOLUTION & MIXED PIXEL
5 Meter resolutionPatalganga, India
5
27-JULY-2011IGARSS,2011-VANCOUVER
SPATIAL RESOLUTION & MIXED PIXEL
6
1. Mixed pixel due to the presence of small, sub-pixel targets within the area it represents .
27-JULY-2011IGARSS,2011-VANCOUVER
SPATIAL RESOLUTION & MIXED PIXEL
7
2. Mixing as a result of the pixel straddling the boundary of discrete thematic classes .
27-JULY-2011IGARSS,2011-VANCOUVER
SPATIAL RESOLUTION & MIXED PIXEL
8
3. Mixing due to gradual transition observed between continuous thematic classes .
27-JULY-2011IGARSS,2011-VANCOUVER
Aral Sea
SPATIAL RESOLUTION & MIXED PIXEL
9
4. Mixing problem due to the contribution of a target (black spot) outside the area represented by a pure but influenced by its point spread function.
So, Mixed Pixels are major concern in satellite image classification !!
27-JULY-2011IGARSS,2011-VANCOUVER
10
When two distinct objects display
similar spectral signatures / Fingerprints
27-JULY-2011IGARSS,2011-VANCOUVER
11
27-JULY-2011IGARSS,2011-VANCOUVER
12
Nature is a Powerful Paradigm
We can learn from nature.
Study of the geographical distribution of biological organisms.
Species migrate between “islands” via flotsam, wind, flying, swimming, …
Habitat Suitability Index (HSI): Some islands are more suitable for habitation than others.
Suitability Index Variables (SIVs): Habitability is related to features such as rainfall, topography, diversity of vegetation, temperature, etc.
27-JULY-2011IGARSS,2011-VANCOUVER
1.Initialize a set of solutions to a problem.
2. Compute “fitness” (HSI) for each solution.
3. Compute S, λ, and μ for each solution.
4. Modify habitats (migration) based on λ, μ.
5. Mutation based on probability.
6. Choose the best candidate & go to step 2 for the next iteration if needed.
13
27-JULY-2011IGARSS,2011-VANCOUVER
14
TERRAIN FEATURES
RADIO SPECTROMETER
SPECTRAL SIGNATURES
BIO-GEOGRAPHY BASEDOPTIMIZATION
DOMAIN EXPERT
1
2
3
4
5
MIXED PIXEL RESOLVED
6
27-JULY-2011IGARSS,2011-VANCOUVER
15
ANALYSING MULTISPECTRAL IMAGE OF ALWAR (RAJASTHAN, INDIA)
False Color Composition Image
27-JULY-2011IGARSS,2011-VANCOUVER
16
27-JULY-2011IGARSS,2011-VANCOUVER
Image Dimension - 476X572 Pixels.
Image’s spectral Bands- LISS-III- Red,Green,Near-Infrared,Middle-Infrared SAR Images- RS1(Low incidence) RS2(High Incidence) DEM(Digital Elevation Model)
Resolution – 25X25 m
17
Satellite & 3-D View of Alwar
27-JULY-2011IGARSS,2011-VANCOUVER
18
DATA SET
27-JULY-2011IGARSS,2011-VANCOUVER
19
RESOLVING THE MIXED PIXEL Satellite Image
1)Identify the Terrain features present in Image (Data set of pure pixels) and the classes of mixed pixel (Data set of Mixed pixels)
Therefore, Each of the mixed pixel corresponds to exactly two of the terrain features.
2)Consider each Terrain feature as Universal Habitat(that comprises of pure pixels). Calculate HSI of each of the Habitat.
[Initially HSI is mean of standard deviation]
3) Take one class of Mixed pixel and transfer each of corresponding mixed pixel to both the Habitats(Terrain feature) to
which it belongs i.e. Immigration & Emigration
C
27-JULY-2011IGARSS,2011-VANCOUVER
20
RESOLVING THE MIXED PIXEL
4) Recalculate the HSI of those two Habitats
If recalculated HSIA<HSIB
Absorb the mixed pixel in Feature A and PPIA ++
Absorb the mixed pixel in Feature B and PPIB++
True False
C
5) Repeat till all the mixed pixels of class taken are resolved
6) Go to step 3 until all classes of mixed pixels are taken and resolved.
O
27-JULY-2011IGARSS,2011-VANCOUVER
PPI-Pure Pixel Index /HSI
21
Water
Vegetation
27-JULY-2011IGARSS,2011-VANCOUVER
JULY,27,2011
22
Water Pixels- 3,5,7,9
Vegetation Pixels-1,2,4,6,8
IGARSS,2011-VANCOUVER
23
•BBO efficiently resolves the mixed pixel & can also be used for other class types.
•BBO mixed pixel resolution algorithm also helps in improving the image classification accuracy and feature extraction.
•Increases the accuracy for the target recognition for air strikes & Defense purpose .
•Can be used for uncovering the enemy camps using the Ariel images.
27-JULY-2011IGARSS,2011-VANCOUVER
24
[1] Ralph W.Kiefer, Thomes M. Lillesand, “Principles of Remote Sensing”,2006.
[2] V.K.Panchal , Sonakshi Gupta, Nitish Gupta, Mandira Monga “Eliciting conflicts in expert’s decision for land use classification”, International Conference on Environment Engineering and Applications, Singapore, pp. 30-33, 2010.
[3] A. Wallace,“The Geographical Distribution of Animals (Two Volumes)”.Boston, MA: Adamant Media Corporation, 2005.
[4] C. Darwin, “The Origin of Species. New York: Gramercy”, 1995.
[5] R. MacArthur and E. Wilson, “The Theory of Biogeography”.Princeton, NJ: Princeton Univ. Press, 1967.
[6] Dan Simon, “Biogeography based optimization”. : IEEE transactions on evolutionary computation, vol. 12, no. 6, December 2008
[7] P. Fisher,”The Pixel: a Snare or a Delusion”, International Journal of Remote Sensing, Vol.18: pp. 679-685, 1997.
27-JULY-2011IGARSS,2011-VANCOUVER
Saturday, February 05,2011
25
NITISH GUPTA ([email protected],[email protected])
V.K.PANCHAL([email protected])
27-JULY-2011IGARSS,2011-VANCOUVER