Upload
helpdesk9532
View
218
Download
0
Embed Size (px)
Citation preview
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
1/10
Pan Sharpening Fusion for Spectral
Change Vector using AWiFS and MODIS
DataDEVESH KHOSLA J.K. SHARMA V.D. MISHRA
STUDENT DIRECTOR SCIENTIST E
RIEIT, S.B.S. Nagar, RIEIT, S.B.S. Nagar, SASE, DRDO
Punjab, India Punjab, India Chandigarh, India,
[email protected] [email protected] [email protected]
Abstract - Change detection is one of the mostimportant applications in the remote sensing society. In change detection, pan sharpening fusion plays an
important role to create a high-resolution dataset. A pan-sharpening method for enhancing satelliteimagery is used as building a relatively high accuracy
and low cost approach for image-based analyses. Inthis paper, Pan sharpening Gram Schmidt technique isapplied for enhancing certain features which are notvisible in either of the single data alone. To perform
the fusion, a low spatial resolution multi-spectralimage of MODIS is fused with higher resolution singleband image of AWiFS of same date using Gram
Schmidt method. Both images are fused for same geographical area and have almost same timeacquisition of lower and middle Himalaya, HimachalPradesh, India. Then improved change vector analysisis used as change detection technique which findsthreshold by double window flexible pace search andalso find the change type discrimination. As resultant
we study the effect of pan sharpening using GramSchmidt on improved change vector analysis afterapplied the proper topographic correction. The
accuracy of land cover change is further comparedwith MODIS data alone.
Keywords: - Pan sharpening, AWiFS and MODIS,Topographic correction, Improved Change VectorAnalysis.
I. IntroductionAs increasing demand of land monitoring and
management of wide area give birth to multi-temporalremote sensing images. These images appear atdifferent temporal and spatial scales and results inchange on earth surface. This change can be measure
by improved change vector analysis (ICVA) onMODIS. But accuracy is not so good due to lack oflesser geometrical and spatial resolution. Spatial andspectral fusion is another approach in which moderatespatial resolution (250m, 500m) multi-spectral
MODIS image is fused with higher spatial resolution(56m) of AWiFS image to enhance change detection.
To take advantage of both geometric and spectralresolution in change detection process, we merge bothproperties of different sensor image in single image ofhaving multispectral properties. This process is calledpan sharpening (fusion). Fused image has topographicinfluence due to rugged topography; therefore
topographic correction is also considered in changedetection analysis. The topographic effect has longbeen recognized as a problem for quantitative analysesof remotely sensed data [1]. This part has become one
of the important image preprocessing steps in the
application of remotely sensed data in mountainousregions. Slope matching technique adopted for terraincorrection is most suitable for Himalayan terrain for
snow cover area [2].A variety of change detection techniques are
available for monitoring land use/land cover changes
[3].The enhancement change detection techniqueshave the advantage of generally being more accuratein identifying areas of spectral change [4]. The change
data are generally created using one of the following:(1) image differencing (2) normalized differencevegetation index (3) change vector analysis (CVA) (4)
principal component analysis and (5) image rationing
[4]. In CVA we use concept of change vector from onedate image to another. The aim of the present paper isto analysis the result of ICVA using (i) pan sharpening
fusion of AWiFS MODIS data along with topographiceffect. (ii) comparative analysis with multi-temporalMODIS data alone.
II. Study areaThe study area is a part of Lower and Middle
Himalaya, India and shown on AWiFS(Advance Wide
Field Sensor) image lies between latitude of 32.26degree to 32.99 degree North and longitude of 77.00degree to 77.49 degree East as shown in the Figure 1.The Lower Himalaya receives the highest snowfall(average 15-20 m) as compared to Middle Himalayan
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 110
mailto:[email protected]:[email protected]:[email protected]:[email protected]8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
2/10
range (12-15m) during the winter period betweenOctober and May. The lower part of the area is
surrounded by forest and tree line exists up to 3100 m.The upper part (Middle Himalaya) is devoid of forest.The average minimum temperature in winter is
generally observed to be -12oC to -15
oC in lower
Himalaya (Pir-Panjal range) and -30o
C to -35o
C inMiddle Himalaya (Greater Himalaya range). The
altitude in the entire study area varies from 1900 m to6500 m with a mean value of 4700 m. The slope in thestudy area varies from 1-86 degree with mean value of28 degree and aspect ranges from 0-360 degree with
mean values of 180 degree. Most of the slopes in thestudy regions are oriented to south aspect.
III. Satellite dataTwo pairs of cloud free satellite images of AWiFS
(Advanced Wide Field Sensor) and MODIS sensor(Moderate Resolution Imaging Spectroradiometer)
acquired on 21 November 2009 (Pre) and 23December 2009(Post) are used in the present work.
The salient specifications of MODIS and AWiFSsensors are given in the Table 1 and Table 2respectively.
IV Image pre-processingA master scene of 56m spatial resolution of
AWiFS(Advance Wide field sensor) of study area is
prepared after rectification with high spatial resolution23m of LISS-III (Linear Imaging self Scanning) with1:50,000 toposheet. This master image was usedfurther to geo-reference MODIS images. All satellite
images were geo-coded to the EVEREST datum byERDAS/Imagine 9.1 (Leica Geosystems GIS and
Mapping LLC) software with a pixel accuracy. The preprocessed topographically uncorrected images of
AWiFS and MODIS are shown in Figure 2 and Figure3 respectively of two different dates.
Figure 1 AWiFS image of the study area
Table 1 Salient Specifications of MODIS Sensor
Spectralbands
Spectralwavelength
Spatialresolution
Quantization(bit)
Radiance Scale(mw/cm2/sr/m)
Radianceoffset
Solar Exoatmostphericspectral Irradiance
(mw/cm2/sr/m)
B1 620-670 250 12 0.0026144 0 160.327
B2 841-876 250 12 0.0009926 0 98.70
B3 459-479 250 12 0.0027612 0 209.071
B4 545-565 250 12 0.0021087 0 186.4
B5 1230-1250 250 12 0.0005568 0 47.6B6 1628-1652 250 12 0.0002572 0 23.8
B7 2105-2155 250 12 0.0000787 0 8.7
Table 2 Salient Specifications of AWiFS Sensor
Spectralbands
Spectralwavelength
Spatialresolution
Quantization(bit)
MaximumRadiance
(mw/cm2/sr/m)
Solar Exoatmostphericspectral Irradiance(mw/cm
2/sr/m)
B2 520-590 56 10 52.34 185.3281
B3 620-680 56 10 40.75 158.042
B4 770-860 56 10 28.425 108.357
B5 1550-1700 56 10 4.645 23.786
N
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 111
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
3/10
Figure 2 Satellite images of 21 November 2009 (a) AWiFS, (b) MODIS
Figure 3 Satellite images of 23 December 2009 (a) AWiFS, (b) MODIS
V Pan sharpeningThe goal of pan-sharpening is to fuse a moderate
spatial resolution multispectral image of MODIS with a
higher resolution single band of multispectral AWiFSimage to obtain an image with high spectral and spatialresolution. Multi-resolution image fusion process(Raster / Combine /Multiresolution Fusion) provides
flexible resolution-enhancement of a multispectralimage using a higher-resolution single band image, a procedure commonly known as pan-sharpening. The
input single band and multispectral band set must havematching spatial extents, and the cell size of the
multispectral bands must be an integer multiple of theband cell size, but no other preprocessing of the imageis required. Different pan sharpening algorithms are
discussed in the literature [5]-[10].Image fusion takes place at three different levels:
pixel, feature, and decision [1]. In pixel-level fusion, anew image is formed whose pixel values are obtained
by combining the pixel values of different images. Thenew image is then used for further processing likeICVA, feature extraction and classification. In feature-
level fusion, the features are extracted from differenttypes of images of the same geographic area. These
(a) (b)
(a) (b)
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 112
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
4/10
extracted features are then used for improved changedetection analysis. In decision-level fusion, the images
are processed separately. The processed information isthen refined by combining the information obtainedfrom different sources . There are various factors to be
considered before performing sharpening [11], [12].
These are The application for which the sharpened data is to be
used. Various satellite image data sets are available andsuitable for sharpening. Co -registration - An important preprocessing step thatshould be done before sharpening is applied to the
multispectral and single band images. The pixel levelaccuracy is achieved in the present paper. Viewing angle of the imagery- If the multispectral and
single band images are taken at different times so thatthe viewing angle is different, registration and pansharpening may not produce a desirable result [13]. Thismay not be required as AWiFS and MODIS (Terra)
have almost same time acquisition. Resampling method and techniques used during
geometric projection, correction, and co-registrationshould be carefully chosen. There is a tradeoff in usingthe different resampling techniques. The co-registeredmultispectral and single band images can be used for
pan sharpening. The gram Schmidt methods, thetechnique introduced by [12] is used in the presentpaper.
Gram-SchmidtSpectral SharpeningIn the Gram-Schmidt (GS) method, as described by
its inventors [14], the spatial resolution of the multi-spectral (MS) image is enhanced by merging the highresolution single band with the low spatial resolution
MS bands. The main steps of the method are as follows:
1. A lower spatial resolution single band is simulated.2. The GS transformation is performed on the simulated
lower spatial resolution single band and the plurality oflower spatial resolution spectral band images. Thesimulated lower spatial resolution single band image isemployed as the first band in the Gram-Schmidt
transformation.3. The statistics of the higher spatial resolution single band is adjusted to match the statistics of the first
transform band resulting from the Gram-Schmidttransformation to produce a modified higher spatialresolution single band image.4. The modified higher spatial resolution single band
image is substituted for the first transform bandresulting from the Gram-Schmidt transformation to
produce a new set of transformed bands.5. The inverse Gram-Schmidt transformation isperformed on the new set of transform bands to producethe enhanced spatial resolution MS image. This
algorithm measures the correlation coefficient, brightness, and contrast between the input and outputbands.
.
Figure 4 Topographic corrected Pan sharpened fused images of AWiFS and MODIS(a) 21 November 2009 (b) 23 December 2009
(a) (b)
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 113
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
5/10
Figure 5 Topographic corrected MODIS images(a) 21 November 2009 (b) 23 December 2009
VI. Radiometrically corrected reflectanceAll geo-referenced satellite images areradiometrically (atmospheric + topographic) corrected.
The image based atmospheric corrections are performedusing the methods proposed in the literature [15]. Thetopographic corrections are than applied using slopematching technique [2] which is reported to be most
suitable for Himalayan terrain. The topographicallycorrected spectral reflectance is estimated using thefollowing equation [2].
= (1)
Where is normalized topographically correctedreflectance, Rij is reflectance on the tilted surfaceand estimated using the model reported in theliterature [17]. Rmax and Rmin are maximum andminimum reflectance of the image, s isillumination on the south aspect, cosi is illumination
(IL) image and estimated using [18] and C is anempirical coefficient. The topographic corrected
fused images of 21 November 2009 and 23December 2009 are shown in Figure 4(a) and Figure4(b) respectively. Topographic corrected MODIS
images of 21 November 2009 and 23 December2009 are shown in Figure 5(a) and Figure 5(b)
respectivelyVII. Improved Change Vector Analysis
A change vector can be described by an angle of
change (vector direction) and a magnitude of changefrom date 1 to date 2 [16], [17]. It is reported [20],Improved change vector analysis is valuable
technique for change detection which includes asemiautomatic method, named Double-WindowFlexible Pace Search (DFPS), which aims atdetermining efficiently the threshold of change
magnitude and a new method of determining changedirection (change category).Which combines a singleimage classification and a minimum-distancecategorization based upon the direction cosines of
the change vector. The improved change vectoranalysis is implemented in this paper as change
detection analysis and results are compared withtopographic inclusion models. There are followingsteps required to perform improved change vectoranalysis. Steps perform for improved change
detection [20] is given in Figure 6.
Change MagnitudeChange Vector is defined as following equation by
[16]
G= H - G = (2)
(a) (b)
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 114
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
6/10
Figure 6 Flow chart of methodology
Where G includes all the change information between
the two dates , for a given pixel by
G = and H= ,
respectively and n is the number of bands, andthe Change magnitude is given by
|G | = (3)
Using above equation, change magnitude was computed.A decision on change is made based on whether the
change magnitude exceeds a specific threshold.
Threshold search using Double-Window Flexible
Pace Search (DFPS)In DFPS method, training sample is selected from
change magnitude image which cover all classes. Samplehaving inner window is an area of interest to find thechange and outer window is used to prevent the
threshold from being too low. As shown in table1.The range of threshold can be selected from max value(P) and minimum value (Q) and divide by any integer
number(R).
= (4)
Where P1 is pace search and R is positive thresholdvalue, which can set manually.
This succession is estimated to fine threshold value.
Succession rate ( ) can be calculated using the
following equation:
= % (5)
Where in equation (5), is number of change pixelsdetected inside an inner training window, is the
number of change pixels detected inside outer training
window incorrectly. It should be noted that outerwindow include one or two pixel only from every side.When highest succession rate is achieved, the iteration is
stopped, then a threshold value that have maximumsuccession rate is applied to an entire change magnitudeimage. The change/no change area from two multi-datefused images (Fig.4) is shown in Figure 7 and change/
no change image from MODIS (Fig.5) is shown inFigure 8
Direction cosineThe direction of cosine is defined as [20]
Cos = , Cos = Cos = (6)
Where X ( , ) is vector, n is number of bands.
Change magnitude ( is calculated using following
equation
(7)
Topographic correction
Land Cover Classification
Raw MODIS satellite image Raw AWiFS Satellite Image
GeocodingGeocoding
Pan Sharpening
Pre Processing (Reflectance)
Change In Magnitude
Threshold (DFPS)
Identified change and no change pixelSeeds Point
Direction Cosine of Change Pixel
Change TypeDiscrimination Based On Classification
Accuracy assessment
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 115
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
7/10
Figure 7 change magnitude of Pan sharpened Figure 8 change magnitude of MODIS images
images(AWiFS + MODIS).
Change type DiscriminationChange type Discrimination can be obtained using
method explained [20], which combines single image
classification with minimum-distance categorizing basedon direction cosines of change vectors. The direction of avector can be described by a series of cosine functions ina multi-dimensional space. This series is called direction
cosines [22]. In this we classified Pre-image bysupervised classification with four classes: (1) Snow, (2)Vegetation, (3) Shadow and (4) Soil. Seed points are
than taken from the reference image which highlight
different class and there mean vector is consider. ThenEuclidean distance of corresponding change is obtained by transplanted these values in direction cosine. Final
Image obtain by minimum distance rule is change image.The minimum distance classified image using Pansharpened and MODIS are shown in the Figure8 and
Figure9 respectively.
VIII. Accuracy assessmentThe accuracy assessment has been carried using
50 random samples (i) determination of threshold of
change magnitude for change/no change classes (ii)from-to change for pan sharpened images (iii) from-tochange for MODIS images alone. To get optimized
threshold we considered highest success rate value asshow in table3 and table4. The threshold of changemagnitude is different for fused and unfused data.Threshold magnitude 70 (table5) is estimated with high
succession rate of 75.18% with pan sharpening ascompared to threshold of 60 (table6) with successionrate of 61.03% for MODIS. These thresholds are
considered for Change / No-Change pixel. The overall
accuracy for change/no change area is estimated to be of96% (Kappa coefficient 0.8339) with pan sharpening ascompared to 94% (Kappa coefficient 0.6431) for
MODIS data alone.The accuracy assessment of from-to change with pansharpening is achieved 92% (Kappa coefficient 0.7481)
as compared to overall accuracy of 88% (Kappacoefficient 0.7273) with MODIS alone and shown in thetable7 and table8 respectively..
(a) (b)
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 116
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
8/10
Figure 8 Minimum Distance Classifiedimage of Fused image
Figure 9 Minimum Distance Classified image ofMODIS
Figure 9 Minimum Distance Classified image of MODIS
Table 3 Results of DFPS with topographic fused image
Range 100-50 Pace 10 Range 75-65 Pace 5 Range 74-66 Pace 2 Range 71-69 Pace 1 Range 70.5-69.5 Pace 0.5
Threshold Success % Threshold Success % Threshold Success % Threshold Success % Threshold
Success %
100 67.20 75 74.53 74 74.06 71 75.18 70.50 75.1890 69.90 70 75.80 72 74.81 70 75.18 70.00 75.18
80 73.30 65 74.43 70 75.18 69 75.18 69.5 75.1870 75.18 68 74.8160 73.68 66 74.06
50 73.68
Table 4 Results of DFPS with topographic MODIS image
Range 90-50 Pace 10 Range 65-55 Pace 5 Range 64-56 Pace 2 Range 61-59 Pace 1 Range 60.5-59.5 Pace 0.5
Threshold Success % Threshold Success % Threshold Success % Threshold Success % Threshold
Success %
90 2.5 65 55.08 64 58.44 61 58.44 60.5 58.480 19.4 60 61.03 62 58.44 60 61.03 60 61.0370 41.55 55 57.15 60 61.03 59 59.74 59.5 59.7
60 61.03 58 58.44
50 51.9 56 57.50
UnclassifiedSoilSnow Vegetation Shadow
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 117
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
9/10
Table 5 Error Matrix using 50 samples for Change/No change Detection with topographic fused image
Referenced Change
Change Pixels No Change Pixels Sum Commission
Error
Classified change Change Pixels 42 1 43 2.30% No Change Pixels 1 6 7 14.20%
Sum 43 7 50Commission Error 2.30% 14.20%
Overall Accuracy= 96 %, Kappa Coefficient = 0.8339
Table 6 Error Matrix using 50 samples for Change/No change Detection with Topographic MODIS image
Referenced Change
Change Pixels No Change Pixels Sum CommissionError
Classified change Change Pixels 44 2 46 2.30% No Change Pixels 1 3 4 14.20%Sum 45 5 50Commission Error 2.30% 14.20%
Overall Accuracy= 94 %, Kappa Coefficient = 0.6431
Table 7 Accuracy Assessment using 50 samples of From-To Change Detection ICVA with Topographic fusedimage
23/12/09 Shadow Snow Soil Vegetation Unclassified Sum
21/11/09 Shadow - - - - - -Snow - 41 4 - - 28Soil - 4 1 - - 19
Vegetation - - - - - 2Unclassified - - - - - 1Sum - 45 5 - - 50
Overall Accuracy =92% , Kappa Coefficient =0.7481
Table 8 Accuracy Assessment using 50 samples of From-To Change Detection ICVA with Topographic modis
image
23/12/09 Shadow Snow Soil Vegetation Unclassified Sum
21/11/09 Shadow - 1 - 1 - 2Snow 5 39 1 2 - 47Soil - - - - - -Vegetation - - - - - -
Unclassified - - - - 1 1Sum 5 40 1 3 - 50
Overall Accuracy =88% , Kappa Coefficient =0.7273
IX. ConclusionA comprehensive analysis of the effects of pan-
sharpened AWiFS and MODIS satellite data onimproved change vector analysis are presented in this
paper. The results are compared with unfused MODISimages. The change-detection maps obtained from the pan-sharpened images in which moderate spatialresolution (250m, 500m) multi-spectral MODIS image
is fused with high spatial resolution (56m) of AWiFS
image to enhance change detection. Analysis of resultsshow that the effect of pan sharpening is a valuable
method for improved change vector with topographiccorrection in which overall accuracy of 92% ( Kappa
coefficient 0.74) is achieved for from-to changeclasses. Which was much greater than MODIS image
alone with topographic correction in which overallaccuracy is 88% (Kappa coefficient 0.72)
AcknowledgementThe authors would like to thank Director Snow
Avalanche Study establishment, Department of DefenceResearch and Development Organization. We are alsothankful to Arun Chaudhary, Scientist, SASE fortechnical discussions.
References
[1] Teillet, P.M., Guindon B., and Goodenough, D.G.,On the slope-aspect correction of multispectral
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 118
8/6/2019 17.IJAEST Vol No 7 Issue No 2 Pan Sharpening Fusion for Spectral Change Vector Using AWiFS and MODIS Data 276
10/10
scanner data, Canadian Journal of Remote Sensing,8:84106,1982.
[2] Mishra V .D., Sharma J .K., Singh K. K., Thakur N.K., and Kumar M., Assessment of differenttopographic corrections in AWiFS satellite imageryof Himalaya terrain J. Earth Syst. Sci,118, No. 1,
pp. 1126, February 2009.[3] Singh, A., Digital Change Detection Techniques
Using Remotely Sensed Data. International Journalof Remote Sensing 10: 989-1003, 1989.
[4] Nelson, R.F., Detecting Forest Canopy Change Due to Insect Activity Using Landsat MSS.Photogrammetric Engineering and Remote Sensing49: 1303-1314, 1983.
[5] Pacifici F., Del Frate F., Solimini C., and EmeryW., An innovative neuralnet method to detecttemporal changes in high-resolution optical satellite imagery, IEEE Trans. Geosci. Remote
Sens., vol. 45, no. 9, pp. 29402952,Sep. 2007.[6] Ranchin T., Aiazzi B., Alparone L., Baronti S., and
Wald L., Image fusion - the ARSIS concept andsome successful implementation schemes, ISPRS J.Photogramm. Remote Sensing, vol. 58, no. 1-2, pp.418, June 2003.
[7] Zhang Y., A new merging method and its spectraland spatial effects, International Journal of RemoteSensing, vol. 20, no. 10, pp.,2003-2014.
[8] Nunez J., Otazu X., Fors O., Prades A., Pala V.,Arbiol R., Multiresolution-Based Image fusionwith additive wavelet decomposition, IEEETransactions on Geoscience and Remote Sensing,
vol. 37, No. 3,pp. 1204-1211, May 1999.[9] King R., Wang J., A wavelet based algorithm for
pansharpening Landsat 7 imagery, Proceedings of
the International Geoscience and Remote Sensingsymposium, vol. 2, pp. 849-851, 2001.
[10]Fanelli A., Leo A., Ferri M., Remote sensingimages data fusion: a wavelet transform approachfor urban analysis, Proceedings of the IEEE/ISPRSjoint workshop on remote sensing and data fusionover urban areas, pp. 112-116, 2001.
[11]T.-M. Tu, S.-C. Su, H.-C. Shyu, and P. S. Huang,Anew look at IHS-like image fusion methods,Information Fusion, vol. 2, no. 3, pp. 177186, Sep.
2001.[12]Chavez P., Jr., S. Sides, J. Anderson, Comparison
of three different methods to merge multiresolution
and multispectral data: Landsat TM and SPOTpanchromatic, Photogrammetric Engineering and
Remote Sensing, vol. 57, no. 3, pp. 295-303, March1991.
[13] Nunez J., Otazu X., Fors O., Prades A., Pala V.,Arbiol R., Multiresolution-Based Image fusion
with additive wavelet decomposition, IEEETransactions on Geoscience and Remote Sensing,
vol. 37, No. 3,pp. 1204-1211, May 1999.[14]Laben C.A., and Brower B.V., Process for
enhancing the spatial resolution of multispectralimagery using pan-sharpening, Tech. Rep. US
Patent # 6,011,875, Eastman Kodak Company,2000.
[15]Mishra,V.D., Sharma,J.K. and Khanna,R.Review oftopographic analysis methods for the western
Himalaya using AWiFS and MODIS satelliteimagery, Annals of Glaciol. 51(54),pp.1-8, 2010.
[16]Jensen, J.R., Introductory Digital ImageProcessing: A Remote Sensing Perspective, SecondEdition, Prentice Hall, Upper Saddle River, NewJersey, 316 p, 1996..
[17] Nichol J., Hang L.K., and Sing W.M.,Empiricalcorrection of low sun angle images in steeplysloping terrain: a slope matching technique; Int. J.Remote Sens. 27(34) 629635M, 2006.
[18]Civco D.L., Topographic normalization of LandsatThematic Mapper digital imager; Photogramm.Eng.Remote Sens. 55 13031309, 1989.
[19]Jensen,J.R., Introductory Digital Image Processing A Remote Sensing Perspective, Second Edition,
Prentice Hall, Upper Saddle River, New Jersey,316 p,1996.
[20]Malila, W.A.,Change vector analysis: an approach for detecting forest changes with Landsat,Proceedings of the 6th Annual Symposium onMachine Processing of Remotely Sensed Data, 03-06 June, Purdue University, West Lafayette,
Indiana, pp. 326335,1980.[21]Jin Chen, Peng Gong, Chunyang He, RuiliangPu,
and Peijun Shi Land-Use/Land-Cover ChangeDetection Using Improved Change-Vector Analysis,Photogrammetric Engineering & Remote SensingVol. 69, No. 4, April 2003, pp. 369379, 2003.
[22]Hoffmann, B., About Vectors, Dover Publications,Inc., New YorkN.Y., 134 p, 1975.
DEVESH KHOSLA* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 7, Issue No. 2, 276 - 285
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 119