Enhanced Normalized Difference Vegetation Index Method based on Stationary Wavelet Transform (SWT) and Singular Value Decomposition (SVD) for Satellite Image Processing

Call for Papers Engineering Research Journal June 2019

Download Full-Text PDF Cite this Publication

Text Only Version

Enhanced Normalized Difference Vegetation Index Method based on Stationary Wavelet Transform (SWT) and Singular Value Decomposition (SVD) for Satellite Image Processing

Sayali Dnyaneshwar Nikam


School of Computer Science & Engineering Sandip University, Nashik, India

Umakant Mandawkar

Ass. Prof.,

School of Computer Science & Engineering Sandip University, Nashik, India

Abstract An enhanced multi-band satellite broadcasting contrast enhancement method dependent on the Stationary Wavelet Transform (SWT) and singular value decomposition (SVD) has been projected for the component extraction of low complexity satellite pictures using standardized distinction vegetation record (NDVI) technique is introduced in this paper. The strategy utilizes multi-spectral remote detecting data method to locate the spectral signature of various items, for example, the vegetation file and land cover grouping presented in the satellite image. The proposed strategy changes over the image into the SWTSVD space and in the wake of normalizing the singular value matrix; the upgraded image is recreated by utilizing inverse SWT. The visual and quantitative outcomes incorporated into this investigation plainly demonstrate the expanded productivity and adaptability of the proposed strategy over the current strategies. The simulation results demonstrate that the improvement based NDVI utilizing SWT SVD strategy is very valuable to distinguish the surface features of the obvious zone which are incredibly useful for municipal planning and the management.

Keywords:- Remote detecting, Normalized difference vegetation index, Stationary Wavelet Transform, Singular Value Decomposition.


The system which is most broadly required in the field of image processing to improve representation of the features with the goal that the surface feature can be accurately extracted by the image enhancement. All in all, raw satellite images have a generally limited scope of brightness values; subsequently, contrast enhancement is often used to improve the multi-band satellite images for better interpretation and visualization [1]. For getting a superior comprehension of the earths environment the multi- spectral remote detecting pictures are productive [1]. Remote detecting is the science and craft of getting data and extricating the highlights as spectral, spatial and temporal about certain objects, region or phenomenon, for example, vegetation, land cover classification, urban zone, farming area and water assets without coming into physical contact of these items [2]. The remote detecting data have numerous application zones including: land cover classification, soil

moisture measurement, forest-type classification, measurement of runny liquid, content of vegetation, snowflake plotting, sea-ice-type cataloguing and oceanography [2, 3]. One of the most common issues, which occur in satellite pictures while catching images with an enormous amount of distance, is dim light and contrast of image. The differentiation is determined by the distinction in the color and brightness of the item with different items in a similar field of view [3]. Essentially, the contrast is created in view of the distinction in luminance which is reflected from two surfaces. In the event that an image has been taken in a dim or a bright situation, the data might be lost in the areas which are too much and consistently dim or bright [4]. The issue is the way the contrast of an image can be enhanced from the input satellite image which has total data but is not visible. There have been a few systems detailed in the writing for contrast examination and improvement of the satellite images, for example, decorrelation extending, straight complexity extends, general histogram evening out (GHE) and discrete wavelet change (DWT) [4]. These procedures are basic and speak to viable lists for complexity improvement. Be that as it may, these procedures were not as proficient as the data laid on the histogram of the image which is completely lost [1].

The multi-spectral remote detecting images convey important integrating spectral and spatial highlights of the items [3]. This paper presents the multi- spectral three-band data of the area and seven-band data of the supplementary area are utilized to compute the level of vegetation present in the image utilizing enhancement and without enhancement procedure. In this manner, the separated component can be accessible to people in general for further examination so as to maintain a tactical distance from any kind of cataclysmic occasions alike floods as well as earthquakes. Remote detecting is an innovation utilized for getting data about an objective through the inspection of data gained from the target at a distance [5]. It is made out of three sections: (a) targets, objects or phenomena in an area; (b) data acquisition through certain instruments; and (c) data analysis by some devices. The derived data have been gained by the utilization of electromagnetic radiation (EMR) in at least one districts of

the electromagnetic range that is reflected or transmitted from the earth's surface. The data exchange is cultivated by the utilization of EMR in its noticeable region and EMR wave is a type of form of energy that reveals its presence by the noticeable impacts its products when it attacks matter [5].

Space Administration of USA (NASA) and The National Aeronautics utilizes seven-band data for highlight extraction and they are called LANDSAT image. The multi- spectral remote-detecting information that implies image of the particular region are acknowledged from the National Remote Detecting Agency and also, they are called INSAT image. It consists of three-band data and the data is practiced by the assistance of these three-groups named close infrared (NIR) band, red band and green band. Each band contains some particular data and with the assistance of these three groups, visible and reflective infrared remote detecting, thermal infrared remote detecting and microwave remote detecting [6]. There are a few files for highlighting vegetation-bearing regions on a remote detecting scene. Normalized difference vegetation index (NDVI) itself a typical and generally utilized list for examination and extraction of vegetation 'present' in the satellite images [7]. It is a significant vegetation list, broadly connected in research on worldwide ecological and climatic change [7]. The vegetation reaction to the earth is sensitive and influences the environmental equalization and atmosphere as well as been observed to be a powerful interruption against catastrophic events. This paper indicates how contrasts between the unmistakable red and NIR groups of an INSAT and LANDSAT images can be utilized to recognize territories containing critical vegetation and other various features [6, 7].

This paper introduces, a new method dependent on the Stationary Wavelet Transform Singular Value Decomposition (SWT SVD) has been proposed for satellite shading image improvement of low- contrast satellite pictures for more accurate feature extraction. The SVD system depends on a theorem from linear algebra which says that a rectangular matrix A, can be separated into the result of three grids, as follows: (a) an orthogonal matrix UA, (b) a diagonal matrix SA and (c) the transpose of an orthogonal matrix VA [4]. The singular value-based image equalization (SVE) strategy depends on leveling the solitary esteem lattice gotten by SVD [1, 8, 9]. Singular Value Decomposition of an image can be translated as a framework, is composed as follows

A = UA A (1)

where UA and VA are orthogonal square matrices identified as hanger and aligner, correspondingly, and the SA matrix holds the sorted singular values on its core diagonal and elementary enhancement arises because of increasing of singular values of the SWT coefficients [8, 9]. Also, the singular value matrix signifies the strength data of image and somewhat modification on the singular standards or values revaluates the strength for the input image. The main benefit of applying SVD for image equalization derives from the statistic that SA contains the intensity data of the image [1, 9]. On account of SVD the proportion of the most noteworthy

solitary estimation of the produced standardized network, with mean zero and variance of one, over a specific image can be given as

= (2)

where SN(µ=0,var=1) is the particular esteem framework of the manufactured power grid. These coefficients can be utilized to recover an equalized image utilizing

Eequalized A = UA(§ ) (3) where Eequalized A is utilized to mean the evened-out

image named A. The adjustment of a image is done to

evacuate the issue of brightening, which is essentially one reason for low-contrast image and blurring [1].

In this work, SWT is connected on the satellite image of each band, for example, NIR band, red band and green band, to extricate better surface features present in the low-contrast satellite pictures. The SWT changes over a spatial area waveform into its constituent frequency components as represented by a lot of coefficients. The way toward reproducing a lot of spatial area samples is known as the inverse singular wavelet transform (ISWT). SWT is utilized to isolate the lower and higher frequency coefficient in two sections. In this paper, SWT is connected on each band to isolate the lower and higher recurrence coefficients of the data image. The SWT features are put away as a cluster where the required features of the image can be determined by SWT recurrence coefficients. Utilizing the underlying couple of qualities, it conveys sensible execution for features extraction of the satellite images. Thus, after backwards of SWT, the upgraded image will be more effective, sharper and have a good contrast.

Numerous analysts have revealed the utilization of NDVI [12] for vegetation checking [13], assessing the crop cover [14], drought monitoring [15, 16] and agricultural drought assessment at national level [17]. In the NDVI strategy, at first various groups are figured from the satellite image and afterward the NDVI technique is connected by its characteristic [12, 13].

In the above context, in this manner another system for improvement is proposed dependent on the joined impact of SWT SVD to separate increasingly exact flexible highlights of satellite images utilizing the NDVI technique.

  1. Overview of SWT SVD and NDVI

    1. Stationary Wavelet Transform (SWT)

      Stationary Wavelet Transform (SWT) algorithm is more popular and is approximation of Stationary Wavelet Transform. The prime difference between SWT with respect to DWT is nothing but the SWT is shift invariant transform algorithm and DWT is shift variant transform algorithm. An issue of shift variance is surviving by either over sampling or eliminating the subsampling stage at individually scale instead of up sampling the channels at each scale in the DWT computation in case of SWT. The removal of subsampling leaves us with redundant data in detail images within each of measures of scales and thus transform is no more of

      orthogonal nature. SWT is superior to many single-phase processes because it is having multiple motion compensation.

      The benefit of DWT is used to reason antiquities in fused images which can be maintained a strategic distance from with the assistance of SWT and its effect will be satisfactory. SWT is more data intensive and more computation intensive than DWT. The SWT can be schematically revealed as in fig.1

      Fig.1.2-D Stationary wavelet transform

      The image dimension should be integral products of 2j, where j signifies number of levels of wavelet decomposition utilized, so as to apply SWT on an image. The symbols hj and hj are correspondingly the low pass as well as high pass wavelet filters used for decomposition at j-th scale. Approximation and detail component of the image input is in correspondence to the outputs of low pass as well as high pass filters

      Four sub images are produced by the SWT at lower scale – A, H, V and D. Sub-images are of similar measure as that of original image. Fig.2. demonstrates 3- level SWT decomposition. The LL, HL, HH and LH sub-bands shown in fig.2 are in correspondence to – A, H, V and D details of input image respectively.

      Fig.2.2-level decomposition of SWT

      With vertical direction the high-pass or detail coefficients characterized the images high frequency information, the low-pass segments hold its low frequency vertical information.

    2. SVD

      SVD can be determined chiefly by the three commonly perfect perspectives. Or then again at the end of the day, SVD is a strategy for changing corresponded factors into a lot of uncorrelated ones that better uncover the different connections among the first data things [20]. In the meantime, it is likewise a technique for distinguishing and requesting the measurements along which data focuses show the most variety [21]. Subsequently, SVD can be viewed as a strategy for data decrease and for the most part for highlight extraction just as for the upgrade of low-differentiate images [21]. Following are the essential thoughts behind SVD: taking a high dimensional, exceedingly factor set of data indicates and decreasing them a lower-dimensional space that uncovered the substructure of the first data all the more plainly and requests them from most variety to the least [20, 21].

    3. NDVI

Spatial and spectral resolutions assume a significant job in portraying remotely detected images. Pixels of a image showed in geometric relationship to each other is turn by spatial and spectral resolutions. NDVI of multispectral remote sensing image depends on belief system of spectral resolutions. Different spectral band give a principal system to understanding element in remotely detected image by varieties in reflectivity of surface material. Utilizing remote sensing data from most other material by prudence of its remarkable in the red ang blue portions of obvious range, its higher green reflectance and particularly, its extremely solid reflectance in the close IR, vegetation can be recognizing. For highlighting area on remote sensing image, there are a few lists. Normally and broadly utilized for the equivalent is NDVI.

The NDVI is usually utilized for estimation of crop health in agricultural applications and detection of vegetation. Inferable from the spatial changeability in soil properties, various areas in a field may require various measures of nitrogen to accomplish high yield [5]. Remotely sensed pictures are categorized by their spatial and spectral resolutions. The terms spatial and resolutions speak to pixels of a image showed in a geometric relationship to each other; and varieties inside pixels as a component of wavelength, individually. NDVI of multi-spectral remote detecting image depends on spatial and spectral resolution. Varieties in reflectivity of surface material crosswise over various spectral groups give a central system to understanding the resolutions in a remotely detected image. Using remote distinguishing information, vegetation can be perceived from the majority of different materials by goodness of its prominent assimilation in the red and blue segments of the noticeable spectrum, its advanced green reflectance and specially, its very robust reflectance in the near-IR [5, 6, 7].


    1. SWTSVD-based enhancement of low-contrast satellite images



      The proposed work of improvement for the satellite image is done in two sections. he first is SVD and is singular value matrix [20]. The SVD contains the illumination data in the image, with the goal that transformation of the singular values will directly change the brightening of the image and the other data present in the image will be same as before [21]. The second most significant part of this strategy is the utilization of SWT. The SWT can be viewed as a discrete- time form of the Fourier-Cosine arrangement. It is firmly related with DFT, a method for changing over a flag into basic recurrence parts. In this manner, SWT can be registered with a FFT. Not at all like DFT, SWT is genuine esteemed and gives a superior guess of a flag with couple of coefficients. In the SWT space, every hugeness condition is reliant on the relating SWT frequency and the recurrence qualities of the significant basic data in the image. Henceforth, isolating the high-recurrence SWT coefficient of each band, for example, NIR band, red band and green band, as appeared in Figs. 1m o and applying the enlightenment improvement in the low-frequency SWT coefficients of each band, it will gather and cover the edge data from conceivable corruption of the multi-band remote sensing satellite images. SWT yield is trailed by applying ISWT and in this manner recreating the last image by utilizing ISWT, the outcome image will be upgraded regarding brightening and it will be sharper with great contrast. In the proposed technique, initially the low-contrast input colored satellite image Ai is managed by GHE to generate Ai . Subsequent to getting this, both of these images are changed by SWT into the lower- frequency SWT coefficient and higher- frequency SWT coefficient. Then, the correction coefficients used for the singular value matrix can be designed by using

      = (4)

      where i {R, G, B}, that is AR, AG, AB and ) is the lower recurrence coefficient singular matrix of the satellite input image, and is the lower- recurrence coefficient singular matrix of the satellite yield output image for the GHE.

      The new satellite image (Di) is examined by

      = § (5)

      = (6)

      Now Di is the lower SWT frequency component of the original image, that is reorganized by applying the ISWT to produce the consequence equalized image Ai as given by (6)

      Step 1: A low-differentiate shaded information satellite picture has been taken for the examination.

      Step 2: Apply equalization technique over the satellite image with the GHE technique.

      Step 3: After the process of equalization, calculate the SWT for the contrast enhancement for each band like NIR band, red band and green band.

      Step 4: Calculate the two variables and from the DCT image. is computed by applying SWT on low-contrast input colored satellite image which is equalized using GHE, whereas is determined directly by applying SWT to low- contrast input colored satellite image. are the lower and higher-frequency coefficients of equalized image using GHE. Whereas are the lower and higher-frequency coefficients of low-contrast input colored satellite image.

      Step 5: After getting and , SVD is applied over calculating the U, S, V and to find the max element in .

      Step 6: Calculate max ) and max with the help of the SVD process.

      Step 7: Calculate § using (4) =

      Step 8: Calculate the new using (8) and (9) = and =

      Step 9: Apply ISWT after getting new Step 10: Equalization of the satellite image.

      Step 11: Separation of NIR band, red band, green band from input low-contrast satellite image, GHE image and proposed SWTSVD image.

      Step 12: NDVI image from input, GHE and proposed SWT SVD image.

      Step 13: Vegetation finding index using NDVI method.

      Step 14: Vegetation finding at 0.2 threshold values of an image from input low-contrast satellite image, GHE image and proposed SWTSVD image.

      Step 15: Creation of false color composite (FCC) image from SWTSVD image for vegetation.

      A complete flowchart routine for the proposed method is shown in Fig.3

      i = ISWT ( i) (7)

      Following advances are to be embraced to talk about the principle computational process of the projected algorithm:

      SWT of (Ai)

      SWT of (Ai)

      Low contrast input colored satellite image

      (Ai) where i {R, G, B} i.e. AR, AG, AB

      Low contrast input colored satellite image

      (Ai) where i {R, G, B} i.e. AR, AG, AB

      Equalized image by GHE

      Equalized image by GHE

      HH HL LH

      SVD, Calculate U, , V and find the max element in

      SWT of (Ai)

      SWT of (Ai)

      LL LH HL

      SVD, Calculate U, , V and find the max element in

      by groups 3 and 4 (obvious red and NIR). The NDVI is determined from these individual estimations as follows

      RNDVI= (7)

      Where 0<NDVI>1 GNDVI= (8)


      where RED is visible red reflectance, and NIR is near-infrared reflectance.

      The wavelength scope for NIR band is (750nm to1300 nm), for red band it is (600nm to 700 nm) and for green band it is (550 nm).

      Calculate using



      max LL ^



      Equalized satellite image

      Equalized satellite image

      max LLAi

      Calculate the new LL

      & LL Ai

      Calculate the new LL

      & LL Ai



      FCC image for vegetation

      FCC image for vegetation


      The NDVI is inspired by the perception vegetation, which is the transformation between the NIR and red band; it ought to be bigger for more noteworthy chlorophyll thickness. It takes the (NIR red) contrast and standardizes it to adjust the impacts of uneven illumination, for example, the shadows of mists or slopes. As it were, on a pixel-by-pixel premise it subtracts the estimation of the red band since the estimation of the NIR band and partitions by their total. Exceptionally low estimations of NDVI (0.1 and beneath) relate to desolate regions of shake, sand, or snow. Moderate qualities speak to bush and prairie (0.2 0.3), though high qualities show mild and tropical rainforests (0.6 0.8). NDVI values go from (21.0 to 1.0), where higher qualities are for green vegetation and low qualities for other normal surface materials. Uncovered soil is implied with NDVI values which are near 0 and water bodies are spoken to with negative NDVI values [5 7].

      using equation (6)

      using equation (6)

      Separation of NIR band, red band, green band from input low contrast satellite image and proposed equalized SWT-SVD image

      Separation of NIR band, red band, green band from input low contrast satellite image and proposed equalized SWT-SVD image

      NDVI image from input and equalized SWT-SVD image

      NDVI image from input and equalized SWT-SVD image

      Vegetation detection using NDVI technique

      Vegetation detection using NDVI technique

      Image of vegetation detection at threshold values from input and improved SWT-SVD image

      Image of vegetation detection at threshold values from input and improved SWT-SVD image

      Remote detecting is the electronic procurement and computerized examination of earth symbolism framework. It is the investigation of determining data about an article from estimations made at a separation from the item, that is, without coming into contact with it. The primary point of remote sensing is to figure the level of various highlights like vegetation area, river, water bodies, and to subsequently make them available to the public for further analysis in order to avoid any sort of natural disasters such as floods.

        1. Finding vegetation in a multi-spectral satellite

          Fig.3.Flowchart for the proposed algorithm


          For the location of the vegetation record from a

    2. Highlight extraction from low-differentiate satellite images utilizing NDVI system

In this area, the NDVI method is utilized for extricating the different features present in the three-band satellite image and futhermore for seven-band data for progressively precise information extraction. The NDVI is a straightforward numerical marker that can be utilized to dissect the remote sensing measurements, from a remote stage and evaluate whether the objective or target being watched contains live green vegetation or not. It is one of the vegetation records that are utilized to quantify the measure of green vegetation [5, 6]. The spectral reflectance contrast among NIR and red is utilized to calculate vegetation and different features by utilizing NDVI system. It is a component of occurrence and reflected light, and the sun is the source of energy. It is a vegetation record characterized

preprocessed multi- spectral remote sensing image, the NDVI procedure needs to separate every single band, which is available in the satellite image. After partition of various band, the NDVI technique is connected by its qualities, for example, vegetation at various NDVI threshold values, for example, 0.2, 0.4, 0.6 and 0.7.

    1. Creation of FCC images

      The alternation of the color for a image, similar to red light for blue, green for red, blue for green, which is called false color image or FCC. Fundamentally, FCC is utilized to recognize some particular features by changing their color. It is helpful for recognizing a specific element present in satellite images. In this paper, vegetation is targeted to detect by utilizing FCC. It is indicated by green color.


In this section, the proposed method has been used for enhancement of several satellite images. Different satellite images are included to demonstrate the usefulness of this algorithm. The performance of this method is measured in terms of the following significant parameters.

Mean (µ) = (x-y) (9)

(a) (b) (c) (d) (e)

Standard Deviation ()=


(f) (g) (h) (i) (j)

Mean (m) is the average of all intensity values. It indicates normal splendor of the picture, though standard deviation is the deviation of the force esteems about mean. It means normal difference of the picture. Here, I(x, y) is the strength value of the pixel (x, y), and (M, N) are the measurements of the image.

Table.1. Comparison of the enhancement results between proposed method with earlier existing techniques

After analysis of the techniques it has been found that the projected technique indicates the better mean (µ) and standard deviation () in comparison with above techniques as show cased in the table.1. The quality of the input image was poor but later smearing the SWT-SVD result is enhanced with orientation of brightness and contrast. The histograms got from of the proposed method are extended in unique range along these lines connoting the improvement interestingly of the yield picture. It is clear from the distributions that the evaluated Gaussian elements of the GHE and the anticipated procedure have implied which are near the perfect mean for the dark dimension extend [with

= 152.056]. Therefore, the perception of the proposed SWT- SVD speaks to the better differentiation just as better splendor with suitable complexity. However, the estimated mean (µ) and standard deviation () in fig. 5(b) of the projected method covers a good range of gray level, and this is the cause of the better illumination and very clear visualization. The enhanced result raise performance level of intensity and contrast of the multispectral images, as a consequence of which, we can easily identify the specific features presented in the satellite images.

(k) (l) (m) (n) (o) Fig.4. Different resulting satellite images using NDVI technique done Decorrelation stretching based enhancement.

(a) (b) (c) (d) (e)

(f) (g) (h) (i) (j)

(k) (l) (m) (n) (o)

Fig.5. Different resulting satellite images using NDVI technique on proposed SWT-SVD enhancement.

[Figs 4(a) and 5(a) show the low contrast satellite images of Paris (France) region. Fig 4(b) shows equalized image by the Decorrelation stretch technique. Fig. 5(b) shows equalized image by the DWT-SVD technique. Fig. 4(c), 4(d) and 4(e) show separated NIR, red and green band from the Decorrelation stretch enhanced image. Fig. 5(c), 5(d) and 5(e) show separated NIR, red and green band from the DWT-SVD enhanced image. Fig. 4(f) is the NDVI image of Decorrelation stretch enhanced image. Fig. 5(f) is the NDVI image of SWT-SVD enhanced image. Fig. 4(g) shows the histogram of the Decorrelation stretch output image. Fig. 5(g) shows histogram of SWT-SVD output image. Fig. 4(h), 4(j), 4(l) and 4(n) are the vegetation index image at 0.2, 0.4, 0.6 &

0.7 threshold value and Fig. 4(i), 4(k), 4(m) and 4(o) are their FCC images which are extracted from Decorrelation stretch enhanced image. Fig. 5(h), 5(j), 5(l) and 5(n) are vegetation index image at 0.2, 0.4, 0.6 & 0.7 threshold value. Fig. 5(i), 5(k), 5(m) and 5(o) are their FCC images which are extracted from SWT-SVD enhanced image].

Table.2. Comparison of the result between input enhanced GHE image and proposed SWT-SVD enhanced image

Sr No



Percentage of vegetation extracted from Decorrelation Stretching Technique

Percentage of vegetation extracted from proposed SWT-SVD enhanced image



15.1730 %

31.9891 %



5.2204 %

24.2460 %



0.8133 %

19.0996 %



0.0027 %

17.0854 %

Table.2 is representing superior presentation of the preprocessing based NDVI technique. It is denoting that low contrast input image is giving very fewer vegetation area at various NDVI threshold value. In the case of low contrast image, the NDVI is incapable to detect exact features due to the pixel overlapping issue during processing of the images. But after enhancement of images, features are obviously visible and vegetation area easily understood by NDVI technique as shown in fig. 5(i), 5(k), 5(m) and 5(o). From the reliability assessment as shown in the table 2, it is initiate that the projected system gives better result in comparison with the existing feature extraction method.


This paper, a new method dependent on SVD SWT space for upgrade of low- contrast satellite images has been projected for feature extraction utilizing the NDVI procedure. The essential improvement happens due to increase in singular values of the SWT coefficients. Execution of this method has been contrasted and the current differentiation upgrade systems like decorrelation extending, LCS, GHE and SVD SWT-based strategies. The investigational outcomes determine that in correlation with the current strategies, the projected method gives better execution regarding contrast (variance) just as brightness (mean) of the enhanced image. Along these lines, this strategy can be viewed as appropriate for improvement of low-contrast satellite image. The NDVI strategy gives superior outcomes for vegetation changing in densities and furthermore for sprinkled vegetation from a multi-spectral remote sensing image. The simulation results demonstrate that the improvement based NDVI, utilizing the SWT SVD method is exceptionally valuable to identify the surface features of the visible region which are incredibly advantageous for city arranging and the management.

Different instruments with the end goal of image preprocessing can be utilized for more advantageous improvement of by and large highlights, differentiate, brilliance, sharpness and interpretability of the remote- detecting information. Normalized different water index can also apply for extraction of water and river bodies. This technique can give better results in case of water resources. The vegetation investigation can be utilized n the circumstance of appalling catastrophic events to give philanthropic guide, harm evaluation and moreover to devise new insurance techniques.


  1. Demirel, H., Ozcinar, C., Anbarjafari, G.: Satellite image contrast enhancement using discrete wavelet transform and singular value decomposition, IEEE Geosci. Remote Sens. Lett., 2010, 7, (2), pp. 333337

  2. Ahmadi, H., Nusrath, A.: Vegetation change detection of Neka River in Iran by using remote-sensing and GIS, J. Geogr. Geol., 2010, 2, (1), pp. 586

  3. Hacihaliloglu, I., Karta, M.: DCT and DWT based image compression in remote sensing images. Proc. IEEE Int. Conf. on Antennas and Propagation Society Int. Symp., 2004, vol. 4, pp. 38563858

  4. Demirel, H., Anbarjafari, G., Jahromi, S.N.M.: Image equalization based on singular value decomposition. Proc. IEEE Int. Conf. on Computer and Information Sciences, 2008, pp. 15

  5. Karaburun, A.: Estimation of C factor for soil erosion modeling using NDVI in Buyukcekmece watershed, Ozean J. Appl. Sci., 2010, 3, pp. 7785

  6. Ramachandra, V.T., Kumar, U.: Geographic resources decision support system for land use, land cover dynamics analysis. Proc.

    FOSS/GRASS users Conf., Bangkok, Thailand, 2004, pp. 1214

  7. Hu, Y., Ban, Y., Zhang, X., Liu, J., Zhuang, D.: Spatialtemporal pattern of GIMMS NDVI and its dynamics in Mongolian plateau. Proc. IEEE Int. Workshop on Earth Observation and Remote Sensing Applications, 2008, pp. 16

  8. Hemes, M.G., Danaher, S., Murray, A.: Characterization of forestry species a comparison using singular value decomposition (SVD) and artificial neural networks (ANN). Proc. IEEE Int. Conf. image Processing and its Applications, 1995, pp. 815819

  9. Murty, S.P., Rajesh, P.K.: A robust digital image watermarking scheme using hybrid DWT-DCT-SVD technique, Int. J. Comput. Sci.

    Network Security, 2010, 10, (1), pp. 185192

  10. Khaleel, A.T.: Enhancement of spatial structure of an image by using texture feature extraction, Al-Rafidain Eng., 2007, 15, (1), pp. 2737

  11. Su Kim, K., Lee, J.M., Lee, K.H.: Blind image watermarking scheme in DWT-SVD domain, IEEE Intell. Inf. Hiding Multimedia Signal Process., 2007, 2, (2), pp. 477480

  12. Nageswara, R.P.P., Shobha, V.S., Ramesh, S.K., Somashekhar, K.R.: Satellite-based assessment of agricultural drought in Karnataka state,

    J. Indian Soc. Remote Sens., 2005, 33, (3), pp. 429434

  13. Tucker, J.C.: Red and photographic infrared linear combinations for monitoring vegetation, Remote Sens. Environ., 1979, 8, pp. 127150

  14. Ayyangar, S.R., Nageswara, R.P.P., Rao, R.K.: Crop cover and phonological information from red and infrared spectral response, J.

    Indian Soc., Remote Sens., 1980, 8, pp. 2329

  15. Nageswara, R.P.P., Rao, R.V.: An approach for agricultural drought monitoring using NAA/AVHRR and Landsat imagery. Proc. Int. Conf.

    Geoscience and Remote Sensing Symp., 1984, vol. 1, pp. 225229

  16. Singh, P.R., Roy, S., Kogan, F.: Vegetation and temperature condition indices from NOAA- AVHRR data for monitoring drought over India region, Int. J. Remote Sens., 2003, 24, (22), pp. 43934402

  17. Thiruvengadachari, S.: National agricultural drought assessment and monitoring system in India: the satellite contribution. Special Publication on Space and Drought Management 1988, pp. 4489

  18. Reeves, R., Kubik, K.: Benefits of hybrid DCT domain image matching, Int. Arch. Photogram. Remote Sens., 2000, 33, pp. 771778

  19. Christopher, J.C., Prabukumar, M., Baskar, A.: Color image enhancement in compressed DCT domain, ICGSTGVIP J., 2010, 10, pp. 3138

  20. Sverdlov, A., Dexter, S., Eskicioglu, A.M.: Robust DCT-SVD domain image watermarking for copyright protection: embedding data in all frequencies. Proc. Int. Conf. on ACM Digital Library on Multimedia and Security, 2004

  21. Ayangar, R.V., Talbar, N.S.: A novel DWT-SVD based watermarking scheme. Proc. Int. Conf. on IEEE Multimedia Computing and Information Technology, 2010, pp. 105108

  22. Hasan Demirel, Cagri Ozcinar, and Gholamreza Anbarjafari: Satellite Image Contrast Enhancement Using Discrete Wavelet Transform and Singular Value Decomposition, 1545-598X/$26.00 2009 IEEE

  23. A. K. Bhandari, A. Kumar and P.K.Padhy.: Satellite Image Processing using Normalized Difference Vegetation Index: A Case Study of Paris City.

Leave a Reply

Your email address will not be published. Required fields are marked *