a flexible scheme for transmission line fault identification using image processing for a secured...

4
Int. Journal of Electrical & Electronics Engg. Vol. 2, Spl. Issue 1 (2015) e-ISSN: 1694-2310 | p-ISSN: 1694-2426 74 NITTTR, Chandigarh EDIT-2015 A Flexible Scheme for Transmission Line Fault Identification Using Image Processing For a Secured Smart Network 1 D.Vijayakumar, 2 V.Malathi 1 Assistant Professor, Department of Electronics and Communication Engineering, LathaMathavan Engineering College, Madurai, India, 2 Professor ,Department of Electrical and Electronics Engineering, Anna University Regional Office, Madurai, Tamil Nadu, India 1 [email protected], 2 [email protected] Abstract:-This paper describes a methodology that aims to find and diagnosing faults in transmission lines exploitation image process technique. The image processing techniques have been widely used to solve problem in process of all areas. In this paper, the methodology conjointly uses a digital image process Wavelet Shrinkage function to fault identification and diagnosis. In other words, the purpose is to extract the faulty image from the source with the separation and the co-ordinates of the transmission lines. The segmentation objective is the image division its set of parts and objects, which distinguishes it among others in the scene, are the key to have an improved result in identification of faults.The experimental results indicate that the proposed method provides promising results and is advantageous both in terms of PSNR and in visual quality. Index TermsImage processing, Fault detection ,Fault diagnosis, Transmission line. I. INTRODUCTION Image is an important way of access to information for people. But noises largely reduce the perceptual quality of images and may result in fatal errors. Image denoising has been a fundamental problem in image processing. The wavelet transform is one of the popular tools in image denoising due to its promising properties for singularity analysis and efficient computational complexity. The noise is occurred in images during the acquisition process, since the intrinsic and thermal fluctuations of acquisition devices. The other reason is only low count photon unruffled by the sensors while comparing others, the signal dependent noise is imperative. It should be unease. Image processing takes part in medical field of the essence. During the disease diagnosis, the consequences of many types of equipment in the medical field are in digital format. There are many prehistoric methods are used for denoisingwhich have its own annoyances. The fundamental undertaking in every sort of picture transforming is discovering an effective picture representation that portrays the noteworthy picture emphasizes in a minimized structure. The main step in order to achieve fault detection and diagnosis is to select a set of inputs whose information is capable to allow the fault identification. This paper uses digital image processing techniques to extract some variables from the tested image. Once all data is collected, it´s necessary to apply digital image processing techniques. These variables are used by the diagnosis tool developed. This strategy is known as bagging and is applied here to improve the power of generalization of the fault detection system . A heuristic is used to determine the optimal number of neurofuzzy networks in the thermovisiondiagnosis.H.m, and MBiswas,[1] stated a generalized picture denoising strategy utilizing neighboring wavelet coef.in Signal,Image and Video Processing techniques. The image processing techniques have been widely used to solve problems in process of all areas [2, 3]. Digital Image Processing consists of a set of techniques used to make transformations in one or more images with the objective to enhance the visual information or scenes analysis to get an automatic perception or recognition from machines [4].Many methods related to image transmission using filtering techniques of multimedia applications over wireless sensor network have been proposed by researchers. Pinar SarisarayBoluk et al. [5] presented two techniques for robust image transmission over wireless sensor networks. The first technique uses watermarking whereas the second technique is based on the Reed Solomon (RS) coding which considers the distortion rate on the image while transmission for wireless sensor networks. Renu Singh et al. [6] proposed wavelet based image compression using BPNN and Lifting based variant wherein optimized compression percentage is arrived using these two adaptive techniques. Pinar SarisarayBoluk, studied image quality distortions occurred due to packet losses using two scenarios, considering watermarked and raw images to improve the Peak-Signal-to-Noise-Ratio (PSNR) rate. In digitalimage processing Zhang Xiao-hong and Liu Gang [7] proposed SPIHT method to reduce the distortion in images. In [8] the authors Wenbing Fan and Jing Chen, Jina Zhen proposed an improved SPIHT algorithm to gain high compression ratio.K. Vishwanath et al. [9] presented image filtering techniques on larger DCT block which speed ups the operation by eliminating certain elements.James R. Carr [10] applied spatial filter theory to kriging for remotely sensed digital images. The method proposed improved image clarity.Buades et al[11] states the Neighborhood filters used for image process and PDE’s. Yu,H.[12] et al mentioned the Image denoising using shrinkage of filter in the wavelet process and joint consensual filter in the dimensional Area.

Upload: ijeee

Post on 17-Aug-2015

32 views

Category:

Engineering


2 download

TRANSCRIPT

Int. Journal of Electrical & Electronics Engg. Vol. 2, Spl. Issue 1 (2015) e-ISSN: 1694-2310 | p-ISSN: 1694-2426

74 NITTTR, Chandigarh EDIT-2015

A Flexible Scheme for Transmission Line FaultIdentification Using Image Processing For a

Secured Smart Network1D.Vijayakumar,2V.Malathi

1Assistant Professor, Department of Electronics and Communication Engineering, LathaMathavan Engineering College,Madurai, India, 2Professor ,Department of Electrical and Electronics Engineering, Anna University Regional Office,

Madurai, Tamil Nadu, [email protected], [email protected]

Abstract:-This paper describes a methodology that aims tofind and diagnosing faults in transmission lines exploitationimage process technique. The image processing techniqueshave been widely used to solve problem in process of allareas. In this paper, the methodology conjointly uses adigital image process Wavelet Shrinkage function to faultidentification and diagnosis. In other words, the purpose isto extract the faulty image from the source with theseparation and the co-ordinates of the transmission lines.The segmentation objective is the image division its set ofparts and objects, which distinguishes it among others in thescene, are the key to have an improved result inidentification of faults.The experimental results indicate thatthe proposed method provides promising results and isadvantageous both in terms of PSNR and in visual quality.

Index Terms—Image processing, Fault detection ,Faultdiagnosis, Transmission line.

I. INTRODUCTIONImage is an important way of access to information forpeople. But noises largely reduce the perceptual quality ofimages and may result in fatal errors. Image denoising hasbeen a fundamental problem in image processing. Thewavelet transform is one of the popular tools in imagedenoising due to its promising properties for singularityanalysis and efficient computational complexity. Thenoise is occurred in images during the acquisition process,since the intrinsic and thermal fluctuations of acquisitiondevices. The other reason is only low count photonunruffled by the sensors while comparing others, thesignal dependent noise is imperative. It should be unease.Image processing takes part in medical field of theessence. During the disease diagnosis, the consequencesof many types of equipment in the medical field are indigital format. There are many prehistoric methods areused for denoisingwhich have its own annoyances. Thefundamental undertaking in every sort of picturetransforming is

discovering an effective picture representation thatportrays the noteworthy picture emphasizes in aminimized structure. The main step in order to achievefault detection and diagnosis is to select a set of inputswhose information is capable to allow the faultidentification. This paper uses digital image processingtechniques to extract some variables from the testedimage. Once all data is collected, it´s necessary to applydigital image processing techniques. These variables are

used by the diagnosis tool developed. This strategy isknown as bagging and is applied here to improve thepower of generalization of the fault detection system . Aheuristic is used to determine the optimal number ofneurofuzzy networks in the thermovisiondiagnosis.H.m,and MBiswas,[1] stated a generalized picture denoisingstrategy utilizing neighboring wavelet coef.inSignal,Image and Video Processing techniques. Theimage processing techniques have been widely used tosolve problems in process of all areas [2, 3]. DigitalImage Processing consists of a set of techniques used tomake transformations in one or more images with theobjective to enhance the visual information or scenesanalysis to get an automatic perception or recognitionfrom machines [4].Many methods related to imagetransmission using filtering techniques of multimediaapplications over wireless sensor network have beenproposed by researchers. Pinar SarisarayBoluk et al. [5]presented two techniques for robust image transmissionover wireless sensor networks. The first technique useswatermarking whereas the second technique is based onthe Reed Solomon (RS) coding which considers thedistortion rate on the image while transmission forwireless sensor networks. Renu Singh et al. [6] proposedwavelet based image compression using BPNN andLifting based variant wherein optimized compressionpercentage is arrived using these two adaptive techniques.Pinar SarisarayBoluk, studied image quality distortionsoccurred due to packet losses using two scenarios,considering watermarked and raw images to improve thePeak-Signal-to-Noise-Ratio (PSNR) rate. In digitalimageprocessing Zhang Xiao-hong and Liu Gang [7] proposedSPIHT method to reduce the distortion in images. In [8]the authors Wenbing Fan and Jing Chen, Jina Zhenproposed an improved SPIHT algorithm to gain highcompression ratio.K. Vishwanath et al. [9] presentedimage filtering techniques on larger DCT block whichspeed ups the operation by eliminating certainelements.James R. Carr [10] applied spatial filter theoryto kriging for remotely sensed digital images. The methodproposed improved image clarity.Buades et al[11] statesthe Neighborhood filters used for image process andPDE’s. Yu,H.[12] et al mentioned the Image denoisingusing shrinkage of filter in the wavelet process and jointconsensual filter in the dimensional Area.

Int. Journal of Electrical & Electronics Engg. Vol. 2, Spl. Issue 1 (2015) e-ISSN: 1694-2310 | p-ISSN: 1694-2426

75 NITTTR, Chandigarh EDIT-2015

II. IMAGE PROCESSING TECHNIQUES

The Digital Image Processing may be divided in thefollowing steps:Image acquisition

Image segmentation.

Image matching using edge detectionNeighShrinkSURETransformation Method.1. Image acquisition

A tested image is used for the process of faultanalysis and diagnosis, for image acquisition. Theparameters of the image are shown in table.

Spatial Resolution(IFOV) 1.3 mrad

Digital Image Enhancement Normal and enhanced

Detector Type

Focal plane array(FPA)Uncooledmicrobolometer

Spectral range7.5 to 13 µm

Focus Automatic or Manual

Electronic zoom function2,4,8 interpolating

Thermal sensitivity@50/60Hz 0.08 c at 30 c

Table.1. Imaging Performance

Fig.1.Process of Identification of faults

2.Wireless networkA WSN (wireless sensor network) is a wireless

network which consists of sensors. Sensors are used tomonitor physical and environmental conditions. Thedevelopment of such networks was originally motivatedby surveillance. The wireless sensor networks are forelectrical systems monitoring.The image will betransmitted from the transmitting end to the receiving endterminal for the analysis of the image.3. Segmentation

Segmentation is the most complex step of imageprocessing system based and it can be made by manyways, depending on the problemcharacteristic and thepurposes to be reached.The segmentation objective is theimage division its set of parts and objects. It’s necessaryto use the whole information available related to theproblem in order to have a successful segmentation. If theobjective is the segmentation of a specific object, its mostmeant characteristics, which distinguishes it among othersin the scene, are the key to have a good result.Thesegmentation objective is the image division its set ofparts and objects. It’s necessary to use the wholeinformation that voltage, current, temperature of thesymmetrical transmission line is given to the inputparameter is set in the image for transmission. It’s evidentthat the acquired image fits the image center to identify itafter the acquisition. The most common tools used at thesegmentation are: Point or Line detection, Edge detection,Gradient operators, Laplacian, Houghlight Simple or

Aptive Threshold, Region Growing and WatershedTransformation. This paper uses gradient operator forachieving better edge preservation. After segmentationthe gradient operator is used for achieving better edgepreservation.4. Image Edge detection

The obtained image will be processed andcompares the image with the input values. Among the keyfeatures of an image i.e. edges, lines, and points, that edgecan be detected from the abrupt change in the gray level.An edge is the border between two different regions. Edgeidentification strategies spot the pixels in the picture thatcompare to the edges of the articles seen in the picture.The result is a binary image with the detected edge pixels.Common algorithms used are Sobel, Canny, Prewitt andLaplacian operators of MATLAB. The gradient iscalculated and the table shows the compared values of theoutput image. Here the gradient based Edge Detectionthat detects the sides by probing for the most andminimum within the differential of theimage.Theimagefunctions can be identified and it iscompared with various PSNR values as shown in thetable.

5. Image DenoisingDespite being more suitable to low bit rate

environments, such as mobile and wireless channels

As a result, it is desirable to remove the noise if possible.It shows that if any fault occurs during the transmissionline the voltage, current and temperature level will get asfaulty values. And the faulty images will get retrieved tothe original values by thresholding the values. Theprinciple behind this is that when noise occurs as faultduring transmission it will be notified the other node .Theretrieving of the fault can be done by Wavelet Shrinkagefunction.

6. Edge Reconstruction for ImagesThe image will be reconstructed by using the

image masking and diagnosis method to retrieve the inputvalues. The input voltage, current and temperature changein values can be monitored by this image processingtechniques. The wavelet-based image compression isadvantageous over the earlier block-based compressiontechniques. Distortion around these edges is perceptuallyobjectionable and cannot be easily avoided if images arerequired to betransmitted at low bit rates. This type ofedge distortion is easily seen in the tested images.

7. Wavelet Shrinkage functionNeighshrinkSURE transformation is one of the

methods in wavelet shrinkage function which is used todiagnosis the faults in transmission line. This waveletthresholding work is reason astute and relies on upon thecoefficients of same area inside the diverse channels, stillas on their oldsters inside the coarser wavelet subband. Anon-excess, orthonormal, wavelet change is beginningconnected to the blunder data ,took after by the (subband-subordinate) vector-esteemed thresholding of individual

Input Image withelectrical parameters

FaultImage

Imagesegmentation

ImageDetection

OriginalImage

Recovery

Int. Journal of Electrical & Electronics Engg. Vol. 2, Spl. Issue 1 (2015) e-ISSN: 1694-2310 | p-ISSN: 1694-2426

NITTTR, Chandigarh EDIT -2015 76

multichannel wavelet coefficient that square measure atlong last conveyed back to the picture area by conversewavelet change. The NeighShrink are efficient imagedenoising algorithms that are based on universal thresholdand discrete wavelet transform. In order to determine theNeighShrink, the optimal threshold and neighbouringwindow size are calculated as,(λs,Ls)=argλ,Lmin SURE(ws,λ,L)………..(1)where λ is optimal threshold, L is neighboring windowsize , s is subbandandsure

(ws,λ,L)= Ns + Σ ||gn(wn)||22 + 2 Σ …...(2)

In equation it is an unbiased estimate of the riskon subband s and L is an odd number and greaterthan1,forexample,3,5,7,9,etc.Ns noisy wavelet coefficientsfrom subband s,

= : , ∈,

into the 1-D vector = : = 1, …gn(wn=

−− ( < )( ℎ )…….…(3)

In the equation (3) gn(wn) is thenth Wavelet coefficient.

InputImage

FaultValuelevel

PSNRValueusing

Neighshrink

MSEvalue

ELAPSEDTIME

1 48.5480 0.9084 10.9161965 36.9973 12.9822 11.701278

10 32.8917 33.4125 10.960715

20 29.342375.6567

11.884871

1 48.5480 0.9084 12.634479

5 36.9973 12.982211.751183

10 32.8917 33.4125 10.833803

20 29.3423 75.6567 10.144114

1 48.3809 0.9440 16.5461695 36.6428 14.0864 16.360380

10 32.6126 35.6301 15.879918

20 29.0225 81.4386 17.639207

Table.2.calculation of PSNR value with elapsed time

III. SIMULATION RESULTSIn order to evaluate the performance of the proposedmethod, the experiment is performed on a representativeset of standard 8 bit gray scale CVG-UGRdatabase, suchas House, Lena, Barbara, Pepper, Boat each of size512x512,256x256 corrupted by simulated additive whiteGaussian noise with a standard deviation equalto10,15,20. Several methods were used to filter the noisyimage. The paper evaluated the performance of proposed

method using the quality measure PSNR which iscalculated as follows.PSNR = 10log ………… (4)Here the performance of proposed method is comparedwith different denoising scheme. Mean Square Error(MSE), which requires two m x n grey-scale, images iandk. Where one of the images is considered as a noisyapproximation of the other and it is defined as: The MSEis defined as:MSE = ∑ ∑ ‖I(i, j) − k(i, j)‖ …(5)

The comparison of PSNR obtained with thesefive different images can be seen in table 2. Table 2comparison isbased on theNeighshrinkSUREtransformation method. As shown in table 2, the PSNR ofimage denoised by the proposed method is obviouslyoutperforms as compared to existing methods. It can beseen that PSNR obtained with the proposed methodisenhanced are compared to existing methods

Fig.2.Input Fig.3.Faulty image

Fig.4.Binary gradient maskFig.5.Dilated gradient mask

Fig.6.Diagnosis Image

IV. CONCLUSIONThis paper presents a method for detection and

diagnostics of failures in transmission line. The inputvalues of the transmission line are injected in the imageand it is transmitted in a network. The obtained imagevalues are processed by the Neigh Shrink SURE function.And if an fault is observed or any noise is occurred in theimage it tends to change the characteristics of the image.Thus the changes can be proceeding by the Neigh

Int. Journal of Electrical & Electronics Engg. Vol. 2, Spl. Issue 1 (2015) e-ISSN: 1694-2310 | p-ISSN: 1694-2426

NITTTR, Chandigarh EDIT -2015 76

multichannel wavelet coefficient that square measure atlong last conveyed back to the picture area by conversewavelet change. The NeighShrink are efficient imagedenoising algorithms that are based on universal thresholdand discrete wavelet transform. In order to determine theNeighShrink, the optimal threshold and neighbouringwindow size are calculated as,(λs,Ls)=argλ,Lmin SURE(ws,λ,L)………..(1)where λ is optimal threshold, L is neighboring windowsize , s is subbandandsure

(ws,λ,L)= Ns + Σ ||gn(wn)||22 + 2 Σ …...(2)

In equation it is an unbiased estimate of the riskon subband s and L is an odd number and greaterthan1,forexample,3,5,7,9,etc.Ns noisy wavelet coefficientsfrom subband s,

= : , ∈,

into the 1-D vector = : = 1, …gn(wn=

−− ( < )( ℎ )…….…(3)

In the equation (3) gn(wn) is thenth Wavelet coefficient.

InputImage

FaultValuelevel

PSNRValueusing

Neighshrink

MSEvalue

ELAPSEDTIME

1 48.5480 0.9084 10.9161965 36.9973 12.9822 11.701278

10 32.8917 33.4125 10.960715

20 29.342375.6567

11.884871

1 48.5480 0.9084 12.634479

5 36.9973 12.982211.751183

10 32.8917 33.4125 10.833803

20 29.3423 75.6567 10.144114

1 48.3809 0.9440 16.5461695 36.6428 14.0864 16.360380

10 32.6126 35.6301 15.879918

20 29.0225 81.4386 17.639207

Table.2.calculation of PSNR value with elapsed time

III. SIMULATION RESULTSIn order to evaluate the performance of the proposedmethod, the experiment is performed on a representativeset of standard 8 bit gray scale CVG-UGRdatabase, suchas House, Lena, Barbara, Pepper, Boat each of size512x512,256x256 corrupted by simulated additive whiteGaussian noise with a standard deviation equalto10,15,20. Several methods were used to filter the noisyimage. The paper evaluated the performance of proposed

method using the quality measure PSNR which iscalculated as follows.PSNR = 10log ………… (4)Here the performance of proposed method is comparedwith different denoising scheme. Mean Square Error(MSE), which requires two m x n grey-scale, images iandk. Where one of the images is considered as a noisyapproximation of the other and it is defined as: The MSEis defined as:MSE = ∑ ∑ ‖I(i, j) − k(i, j)‖ …(5)

The comparison of PSNR obtained with thesefive different images can be seen in table 2. Table 2comparison isbased on theNeighshrinkSUREtransformation method. As shown in table 2, the PSNR ofimage denoised by the proposed method is obviouslyoutperforms as compared to existing methods. It can beseen that PSNR obtained with the proposed methodisenhanced are compared to existing methods

Fig.2.Input Fig.3.Faulty image

Fig.4.Binary gradient maskFig.5.Dilated gradient mask

Fig.6.Diagnosis Image

IV. CONCLUSIONThis paper presents a method for detection and

diagnostics of failures in transmission line. The inputvalues of the transmission line are injected in the imageand it is transmitted in a network. The obtained imagevalues are processed by the Neigh Shrink SURE function.And if an fault is observed or any noise is occurred in theimage it tends to change the characteristics of the image.Thus the changes can be proceeding by the Neigh

Int. Journal of Electrical & Electronics Engg. Vol. 2, Spl. Issue 1 (2015) e-ISSN: 1694-2310 | p-ISSN: 1694-2426

NITTTR, Chandigarh EDIT -2015 76

multichannel wavelet coefficient that square measure atlong last conveyed back to the picture area by conversewavelet change. The NeighShrink are efficient imagedenoising algorithms that are based on universal thresholdand discrete wavelet transform. In order to determine theNeighShrink, the optimal threshold and neighbouringwindow size are calculated as,(λs,Ls)=argλ,Lmin SURE(ws,λ,L)………..(1)where λ is optimal threshold, L is neighboring windowsize , s is subbandandsure

(ws,λ,L)= Ns + Σ ||gn(wn)||22 + 2 Σ …...(2)

In equation it is an unbiased estimate of the riskon subband s and L is an odd number and greaterthan1,forexample,3,5,7,9,etc.Ns noisy wavelet coefficientsfrom subband s,

= : , ∈,

into the 1-D vector = : = 1, …gn(wn=

−− ( < )( ℎ )…….…(3)

In the equation (3) gn(wn) is thenth Wavelet coefficient.

InputImage

FaultValuelevel

PSNRValueusing

Neighshrink

MSEvalue

ELAPSEDTIME

1 48.5480 0.9084 10.9161965 36.9973 12.9822 11.701278

10 32.8917 33.4125 10.960715

20 29.342375.6567

11.884871

1 48.5480 0.9084 12.634479

5 36.9973 12.982211.751183

10 32.8917 33.4125 10.833803

20 29.3423 75.6567 10.144114

1 48.3809 0.9440 16.5461695 36.6428 14.0864 16.360380

10 32.6126 35.6301 15.879918

20 29.0225 81.4386 17.639207

Table.2.calculation of PSNR value with elapsed time

III. SIMULATION RESULTSIn order to evaluate the performance of the proposedmethod, the experiment is performed on a representativeset of standard 8 bit gray scale CVG-UGRdatabase, suchas House, Lena, Barbara, Pepper, Boat each of size512x512,256x256 corrupted by simulated additive whiteGaussian noise with a standard deviation equalto10,15,20. Several methods were used to filter the noisyimage. The paper evaluated the performance of proposed

method using the quality measure PSNR which iscalculated as follows.PSNR = 10log ………… (4)Here the performance of proposed method is comparedwith different denoising scheme. Mean Square Error(MSE), which requires two m x n grey-scale, images iandk. Where one of the images is considered as a noisyapproximation of the other and it is defined as: The MSEis defined as:MSE = ∑ ∑ ‖I(i, j) − k(i, j)‖ …(5)

The comparison of PSNR obtained with thesefive different images can be seen in table 2. Table 2comparison isbased on theNeighshrinkSUREtransformation method. As shown in table 2, the PSNR ofimage denoised by the proposed method is obviouslyoutperforms as compared to existing methods. It can beseen that PSNR obtained with the proposed methodisenhanced are compared to existing methods

Fig.2.Input Fig.3.Faulty image

Fig.4.Binary gradient maskFig.5.Dilated gradient mask

Fig.6.Diagnosis Image

IV. CONCLUSIONThis paper presents a method for detection and

diagnostics of failures in transmission line. The inputvalues of the transmission line are injected in the imageand it is transmitted in a network. The obtained imagevalues are processed by the Neigh Shrink SURE function.And if an fault is observed or any noise is occurred in theimage it tends to change the characteristics of the image.Thus the changes can be proceeding by the Neigh

Int. Journal of Electrical & Electronics Engg. Vol. 2, Spl. Issue 1 (2015) e-ISSN: 1694-2310 | p-ISSN: 1694-2426

77 NITTTR, Chandigarh EDIT-2015

ShrinkSURE function and the original values of the inputare obtained. Thus it ensures that the fault can be detectedand it is diagnosed.The diagnostics tool implementedshowed itself as a powerful tool to identify the fault.

REFERENCES[1] Hari Om, and MantoshBiswas, “A generlz. Image denoising mtd.Using neighbr. waveletcoef.,” Signal,Image and Video ProcessingSpringer -SViP March, 2013[2] P. Sollich and A. Krogh, “Learning with ensembles: how over-fittingcan be useful”, In D. S. Touretzky, M. C. Mozer, and M. E. Hasselmoeds., Advances in Neural Information Processing Systems 8, Denver,CO, MIT Press, Cambridge, MA,pp.190-196, 1996.[3] L. K. Hansen and P. Salamon, “Neural network ensm. , IEEE Trans.Pattern Analys.and Mc. Int., vol.12(10):pp.993-1001, 1990.[4] William K. Pratt. Digital Image Processing. John Wiley e Sons,INC., 2 edition, 1991.[5] Pinar SarisarayBoluk, SebnemBaydere, A. EmreHarmanci, “RobustImage Transmission Over Wireless Sensor Networks “, MobileNetworks and Applications, ACM , Vol. 16 .Apr 2011

[6] Renu Singh, SwanirbharMajumder, U. Bhattacharjee, A. DinamaniSingh,” BPNN and Lifting Wavelet Based Image Compression”,Information and Communication Technologies, Vol. 101.,Springer 2010[7] Zhang Xiao-hong, Liu Gg.,” Res. of the SPIHT Comp.Based onWavelet Int.pol.Matg. Image”, Communications in Computer andInformation Science, Vol. 225,Springer 2011 .[8] Wenbing Fan, Jing Chen, Jina Zhen,” SPIHT Algorithm Based onFast Lifting Wavelet Transform in Image Compression “, ComputationalIntelligence and Security,Vol. 3802, Springer 2005.[9] K. Viswanath, Jayanta Mukherjee, P.K. Biswas,” Image filtering inthe block DCT domain using symmetric convolution”, ACM Vol. 22,Feb 2011[10] James R. Carr, “Application of spatial filter theory to kriging”,Mathematical Geology, Springer Nov 1990 Vol. 22 .[11] A. Buades, B. Coll, and J. Morel, “Neighborhood filters andPDE’s,” Numer. Math., vol. 105, pp. 1–34, 2006.[12] Yu,H., Zhao, L.,Wang, H., “ Image denoising using TS filter inthewavelet domain and jt.bil.filter in the sptlDmin”. IEEE Tr.Im.Procs.Vol.19(10),pp. 2364–2369,2009.