IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

International Journal of Research in Information Technology (IJRIT)

www.ijrit.com

ISSN 2001-5569

Reversible Information Hiding in Encrypted Images by Reserving Room before Encryption 1

K.S.Parimala , C.Preethi

2

1

Professor and HOD (Computer Science dept.,), Bharat institute of technology & science for women, Mangalpally, Ibrahimpatnam(M), RR Dt. – 501510. [email protected] 2

Bharat institute of technology & science for women, Computer Science and Engineering, Mangalpally, Ibrahimpatnam(M), RR Dt.– 501510. [email protected]

Abstract: Lately, more and more attention is undergone punishment for to reversible facts skin, leather (RDH) in encrypted images, since it maintains the very good, of highest quality property that the first form cover can be losslessly got loss back in law after fixed knowledge for computers is got from while safe-keeping the image what is in secretly. All earlier methods fix knowledge for computers by reversibly vacating room from the encrypted images, which may be person to some errors on knowledge for computers extraction and/or image placing back. In this paper, we make an offer a new way by keeping back room before encryption with an old and wise RDH algorithm, and thus it is simple, not hard for the facts hider to reversibly fix facts in the encrypted image. The made an offer way can get done true reversibility, that is, facts extraction and image got over a disease are free of any error. experiments make clear to that this fiction story way can fix more than times as greatly sized onboard instruments for the same image quality as the earlier ways of doing, such as for PSNR = 40 dB.

1 Introduction Reversible facts skin, leather (RDH) in images is a way of doing, by which the first form cover can be losslessly got loss back in law after the fixed note is got from. This important way of doing is widely used in medical picturing, military picturing and law forensics, where no distortion of the first form cover is let. Since first presented, RDH has get attraction much research interest. In theoretical point of view, Kalker and Willems got started a rate-distortion design to be copied for RDH, through which they proved the rate-distortion bounds of RDH for memoryless covers and made an offer a recursive general rule of behavior of a group making which, however, does not move near the joined. Zhang et Al, got better the recursive general rule of behavior of a group making for based on covers and proved that this making can get done the rate-distortion joined as long as the forced together algorithm reaches entropy, which gets started the being equal between facts forced together and RDH for based on covers. In useful point of view, many RDH techniques have came out of in nearby years. Fridrich et Al. made a general framework for RDH. By first getting from compressible features of first form cover and then getting forced together them losslessly, let free space can be kept safe for getting fixed helping facts. A more pleasing to all way is based on point or amount unlike expansion (DE) , in which the point or amount unlike of each bit of picture group is got

K.S.Parimala, IJRIT

372

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

wider (greater), e.g., multiplied 7 by 2, and thus the least important bits (LSBs) of the point or amount unlike are allzero and can be used for getting fixed notes. Another making statement of undertaking secret design for RDH is histogram shift (Hs) , in which space is kept safe for facts getting fixed by changing the boxes of histogram of gray values. The state-of-art methods usually has at need DE or Hs to residuals of the image, e.g., the predicted errors, to get done better operation. With in connection with to making ready secretly for images, encryption is a working well and pleasing to all means as it gets changed into the first form and purposeful. What is in to not able to see clearly one. Although few RDH techniques in encrypted images have been made public still, there are some making statement of undertaking applications if RDH can be sent in name for to encrypted images. In , Hwang et Al. gave support a reputation-based trust-management design gave greater value to with facts coloring (a way of getting fixed facts into covers) and software watermarking, in which facts encryption and coloring offer possible states for supporting the what is in owners right not to be public and knowledge for computers true, good nature. clearly, the cloud public organization giver has no right to present fixed distortion during knowledge for computers coloring into encrypted facts. In this way, a reversible facts coloring way of doing based on encrypted facts is had a better opinion of. take as probable a medical image knowledge-base is stored in a facts inside, and a computer in the facts inside can fix system of naming into an encrypted account of a medical image through a RDH expert way of art and so on. With the system of naming, the computer can manage the image or make certain of its true, good nature without having the knowledge of the first form What is in, and thus the persons getting care right not to be public is kept safe (out of danger). On the other hand, a doctor, having the cryptographic key, can decrypt and make like new, healthy, normal the image in a reversible ways for the purpose of further working out. Some attempts on RDH in encrypted images have beenmade. In , Zhang separated the encrypted image into several gets in the way. By flipping LSBs of the half of bit of picture in each solid mass, room can be vacated for the fixed bit. The facts extraction and image got over a disease go on (forward) by discovering which part has been let chance make decision in one solid mass. This process can be realizedwith the help of spatial connection in decrypted image. Hong et Al. turned for the better Zhangs way at the decoder side by further using persons wrongly the spatial connection using a different rough statement equation and side match way of doing to get done much lower error rate. These two methods said-about above get support from on spatial connection of uncommon, noted image to get out facts. That is, the encrypted image should be decrypted first before facts extraction. To separate the knowledge for computers extraction from image decryption, Zhang put out space for knowledge for computers getting fixed supporters the idea of making shorter encrypted images,forced together of encrypted knowledge for computers can be put clearly as starting point coding with side information at the decoder , in which the of a certain sort way is to produce the make shorter knowledge for computers in lossless ways by making use of the representative group of signs of parity-check matrix of narrow way put into signs. The way in make shorter the encrypted LSBs to give up living in room for added facts by having experience representative group of signs of a parity-check matrix, and the side information used at the radio side is also the spatial connection of decrypted images. All the three methods attempt to give up living in room from the encrypted images going straight to something. However, since the entropy of encrypted images has been made greatest amount, these techniques can only get done small onboard instruments , or produce marked image with poor quality for greatly sized onboard instruments and all of them are person to some error rates on facts extraction and/or image placing back. Although the methods in , can put out waste (from body) errors by errorcorrecting put into signs, the clear onboard instruments will be further destructed. In the present paper we make an offer a new way for RDH in encrypted images for which we do not give up living in room after encryption as done in but kept back room before encryption. In the made an offer way we first with nothing in out room by getting fixed LSBs of some bit of picture into other bit of picture with an old and wise RDH

K.S.Parimala, IJRIT

373

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

way and then encrypt the image so the positions of these LSBs in the encrypted image can be used to fix facts Not only does the made an offer way separate knowledge for computers extraction from image decryption but also gets done very good, of highest quality operation in two different prospects. True reversibility is got MONEY for that is knowledge for computers extraction and image got over a disease are free of any error. For given getting fixed rates the PSNRs of decrypted image having in it the fixed knowledge for computers are importantly got better and for the pleasing PSNR the range of getting fixed rates is greatly made greater.

2 Previous Arts The methods made an offer in can be made a short account as the framework vacating room after encryption VRAE as pictured in Fig 1a. In this framework a What is in owner encrypts the uncommon, noted image using a quality example cipher with an encryption key. After producing the encrypted image the content owner hands over it to a knowledge for computers hider e.g., a knowledge-base manager and the facts hider can fix some helping facts into the encrypted image by losslessly vacating some room according to a knowledge for computers putting out of the way key. Then a radio possibly the What is in owner himself or a given authority third meeting of friends can get out the fixed knowledge for computers with the knowledge for computers putting out of the way key and further get back the uncommon, noted image from the encrypted account according to the encryption key. In all methods of the encrypted bit gray scale images are produced by encrypting every bit planes with a small river cipher. The way in parts the encrypted image into a number of non overlapping gets in the way sized by an an each solid mass is used to keep one added bit. To do this bit of picture in each solid mass are false as by chance separated into two groups s1 and s2 and according to a knowledge for computers putting out of the way key. If the added bit to be fixed is 0 make sudden move the 3 LSBs of each encrypted bit of picture in s1 otherwise make sudden move the encrypted LSBs of bit of picture in s2. For facts extraction and image got over a disease the radio makes sudden move all the three LSBs of bit of picture in s1 to form a new decrypted solid mass and lets chance make decision all the three LSBs of bit of picture in s2 to form another new solid mass one of them will be decrypted to the first form solid mass needing payment to spatial connection in natural images uncommon, noted get in the way of is took as true to be much smoother than got in the way solid mass and fixed bit can be got from rightly. However there is a danger of make of no effect of bit extraction and image got over a disease when separated solid mass is relatively small e.g., or has much in very small grains detailed textures. Hong et Al made lower, less the error rate of Zhang smethod by fully making use of the bit of picture in designing the smoothness of each solid mass and using side match. The extraction and get loss back in law of gets in the way are did according to the sloping down order of the complete smoothness point or amount unlike between two going up for position gets in the way and got loss back in law gets in the way can further be used to value the smoothness of unrecovered gets in the way which is has relation to as side match. Zhang s way in false as by chance permuted and separated encrypted image into a number of groups with size of the P LSB-planes of each group are make shorter with a parity check matrix and the vacated room is used to fix facts For example be the sign of the bit of picture of one group by x1,…..xL and its encrypted P LSB-planes by that is chiefly of P bits. The facts hider produces a parity check matrix G sized (P.L-S)×P.L and got forced together as its representative group of signs such that s=G.c. Because the length of is (P.L-S) bits are ready (to be used) for knowledge for computers rooms at the radio side the P most important bits MSB of bit of picture are got by decryption directly. The radio then estimates xi(1≤  ≤ r) by the MSBs of near bit of picture and gets a put a value

K.S.Parimala, IJRIT

374

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

on account of detailed by on the other hand the radio tests each guide being the property of to the coset Ω(s) of representative group of signs where Ω(s)={u | G.u=s}. From each guide of Ω(s) the radio can get a put back to earlier position account of and select the one most similar to the put a value on account c′ as the put back to earlier position LSBs.

3 Proposed Method Since losslessly vacating room from the encrypted images is relatively hard and sometimes inefficient why are we still so took up mind (attention) completely to discover fiction story RDH techniques working directly for encrypted images. If we opposite the order of encryption and vacating room i.e., keeping back room before to image encryption at what is in owner side the RDH tasks in encrypted images would be more natural and much more comfortable which leads us to the fiction story framework keeping back room before encryption RRBE. As given view in Fig 1b the What is in owner first reserves enough space on uncommon, noted image and then gets changed into the image into its encrypted account with the encryption key. Now the facts fix ding process in encrypted images is inherently reversible for the facts hider only needs to give space facts into the let free space earlier put out. The facts extraction and image got over a disease are the same to that of framework VRAE clearly quality example RDH algorithms are the high-purpose operator for keeping back room before encryption and can be easily sent in name for to framework RRBE to get done better operation made a comparison with techniques from framework VRAE. This is because in this new framework we move after the as generally done idea that first losslessly got forced together the redundant image happy e.g., using very good, of highest quality RDH techniques and then encrypts it with respect to safe-keeping right not to be public. The one after we elaborate a useful way based on the framework RRBE which primarily is chiefly of four stages stage of encrypted image data putting out of the way in encrypted image data extraction and image got over a disease note that the keeping back operation we take up in the made an offer way is an old and wise RDH move near.

Fig. 1. Framework: “vacating room after encryption (VRAE)” versus framework: “reserving room before encryption (RRBE).” (Dashed line in (a) states that the need of data hiding key in image recovery varies in different practical methods). (a) Framework VRAE. (b) Framework RRBE. A. Generation of Encrypted Image

K.S.Parimala, IJRIT

375

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

Actually to make the encrypted image the first stage can be separated into three steps image structure separating self reversible getting fixed moved after by image encryption. At the starting image structure separating step makes a division uncommon, noted image into two parts A and B then the LSBs of A are reversibly fixed into B with a quality example RDH algorithm so that LSBs of A can be used for ready to do notes at last encrypt the made orderly again image to produce its last account. 1) Image Partition: The operator here for keeping back room before encryption is a quality example RDH way of doing so the making person do by wounding feelings of image structure separating is to make a smoother area on which quality example RDH algorithms such as can get done better operation. To do that without loss of generality take to be true the uncommon, noted image is a 8 bits gray scale image with its size M×N and bit of picture. First the What is in owner gets out from the uncommon, noted image along the lines several partly covering gets in the way whose number is strong of purpose by the size to be fixed notes detailed by in detail every get in the way of is chiefly of lines where M= [l/N] and the number of gets in the way can be worked out through n=M-m+1. An important point here is that each solid mass is partly covered by pervious and or subsequential gets in the way along the lines. For each solid mass make statement of the sense of words a purpose, use to measure its first order smoothness.

Fig. 2. Illustration of image partition and embedding process. Higher gives the story of to blockswhich have within relatively more complex textures. The what is in owner therefore selects the one solid mass with the highest to be and puts it to the front of the image got joined together by the rest part B with fewer textured areas as given view in Fig 2. The above discussion unquestioning is dependent on the fact that only single LSB plane of A is recorded It is straightforward that the What is in owner can also fix two or more LSB planes of A into B which leads to half or more than half copies of smaller size in size of A. However the doing a play of A in terms of PSNR after facts getting fixed in the second stage drops importantly with growing bit planes made use of as an outcome of that in this paper we research situations that at most three LSB planes of A are given work and come to a

K.S.Parimala, IJRIT

376

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

decision about the number of bit plane with inω connection with to different onboard instruments as a test in the next part. 2) Self Reversible Embedding: The end, purpose of self reversible getting fixed is to fix the LSB planes of A into B by using old and wise RDH algorithms. For example, image we make simpler the way in to put examples on view the process of self getting fixed note that this step does not have belief in on any special RDH algorithm. Bit of picture in the rest of image are first sorted into two puts white bit of picture with its pointers i and j making free from doubt (i + j) mod 2=0 young person of up to day dress and black bit of picture whose pointers meet (i+j) mod 2=1 young person of up to day dress as given view in Fig 2. Then each white bit of picture Bi,j, is put a value on by the interpolation value got with the four black bit of picture all round, nearby it as follows. B′i,j = ω1Bi-1,j+ω2Bi+1,j+ ω3Bi,,j-1+ ω4Bi,,j+1,

(2)

Where the weight wi, 1≤ i ≤4, is strong of purpose by the same way as made an offer in, the putting a value on error is worked out via eij = Bij – B′ij and then some facts can be fixed into the putting a value on error order with histogram shift which will be described later. After that we further work out the putting a value on errors of black bit of picture with the help of all round, nearby white bit of picture that may have been made an adjustment. Then another putting a value on error order is produced which can give space notes as well in addition we can also instrument many-level getting fixed design by giving thought to as the made an adjustment b as uncommon, noted one when needed. In short account to make use of all bit of picture of B two putting a value on error orders are made for getting fixed notes in every single level getting fixed process. By bidirectional histogram shift some notes can be fixed on each error order. That is first separate the histogram of putting a value on errors into two parts i.e., the left part and the right part and look for the highest point in each part detailed by LM and RM separately for of a certain sort images LM = ‫־‬1 and RM =0 in addition look for the zero point in each part detailed by LN and RN. To fix notes into positions with a putting a value on error that is equal to RM shift all error values between RM + 1 and RN -1 with one step toward right and then we can represent the bit 0 with RM and the bit 1 with RM + 1. The getting fixed process in the left part is similar except that the changing direction is left and the shift is got money for by taking away from the being like (in some way) bit of picture values. Take as probable we should give effect to the getting fixed design X times to give space added facts. In the earlier x1 single level getting fixed rounds highest point points of two error orders are selected and put to use to fix notes as above said-about when it comes to the xth single level getting fixed only a small part of notes is left to be fixed so it is unadvisable to give space such little facts at the money used, needed, for something of changing all error values between highest point points and their being like (in some way) zero points. To amount with this question under discussion we can either make use of only part of error orders which has enough highest point points to fix the still in the same way notes while going away from the rest error orders unchanged or discover two right points detailed by LP and RP whose addition is larger however nearest to the size of still in the same way notes. By changing error values between LP,RP and their being like (in some way) zero points notes can be fixed into LP and RP instead of highest point points . Fig. 3 pictures the idea of selecting right points. generally speaking, two answers can profit importantly getting better in terms of PSNR when the length of knowledge for computers is relatively short, i.e., when x=1. and the being higher, greater of one answer over the other depends highly on statistics of natural image itself which will be had a discussion about in the next part. The same with other RDH algorithms, overflow/underflow hard question comes to mind when natural division line bit of picture change from 255 to 256 or from 0 to -1. To keep from it, we only fix facts into putting a value on error with its being like (in some way) bit of picture valued from 1 to 254. However, with more than one possible sense still get up when nonboundary bit of picture are changed from 1 to 0 or from 254 to 255 during the getting fixed process. These made come into existence division line bit of picture in the getting fixed process are formed as pseudo-boundary bit of picture, for this reason, a division line map is introduced to say to whether division line bit of picture in marked image are natural or false in getting from process. It is a based on order with bit 0 for natural division line bit of picture, bit 1 for pseudo-boundary bit of picture. Since putting a value on errors of marginal area

K.S.Parimala, IJRIT

377

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

of B can not be worked out via (2), to make the best use of B we select its marginal area made clear in Fig. 2 to place the division line map, and use LSB putting in place of to fix it. The first form LSBs of marginal area is got

(3)

Fig. 3. Selection of proper points. (a) original histogram, (b) shifted histogram. (In this figure, length of messages is 1000 bits, and .)

together with notes, i.e., LSB-planes of , and reversibly fixed into . In most examples, even with a greatly sized getting fixed rate, the length of division line map is very short; in this way, the marginal area of B is enough to give space it. meanwhile, several parameters such as LN ,RN ,LM ,RM ,LP ,RP.payloads fixed into the putting a value on errors of black bit of picture Rb,total getting fixed rounds x, start line SR and end street fight, trouble ER of A in uncommon, noted image, are fixed into marginal area in a similar way. These parameters play an important undertakings in knowledge for computers extraction and image got over a disease process. 3) Image Encryption: After rearranged self-embedded image, denoted by , is generated, we can encrypts X to construct the encrypted image, denoted by E.With a stream cipher, the encryption version of X is easily obtained. For example, a gray value Xi,j ranging from 0 to 255 can be represented by 8 bits, , such that Xi,j(0), Xi,j(1),….., Xi,j(7),such that The encrypted bits Ei,j(k) can be calculated through exclusive- or operation Ei,j(k) =Xi,j(k) ⨁ ri,j(k),

(4)

where ri,j(k) is generated via a standard stream cipher determined by the encryption key. Finally, we embed 10 bits information into LSBs of first 10 pixels in encrypted version of A to tell data hider the number of rows and the number of bit-planes he can embed information into. Note that after image encryption, the data hider or a third party

K.S.Parimala, IJRIT

378

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

can not access the content of original image without the encryption key, thus privacy of the content owner being protected.

B. Data Hiding in Encrypted Images Once the facts hider gets the encrypted image , he can fix some facts into it, although he does not get way in to the uncommon, noted image. The getting fixed process starts with giving position of the encrypted account of A, detailed by AE. Since AE has been made orderly again to the top of , it is effortless for the facts hider to read 10 bits information in LSBs of first 10 encrypted bit of picture. After having knowledge of how many bit-planes and lines of bit of picture he can modify, the facts hider simply takes up LSB putting in place of to use for another the ready (to be used) bit-planes with added facts, at last, the facts hider puts a ticket giving name (joined to clothing) supporters m to point out the end position of getting fixed process and further encrypts m according to the knowledge for computers putting out of the way key to put clearly marked encrypted image detailed by E'. Anyone who does not have as owner the knowledge for computers putting out of the way key could not get out the added facts. C. Data Extraction and Image Recovery Since facts extraction is completely independent from image decryption, the order of them suggests two different useful requests. 1) Case 1: Getting from facts From Encrypted images: To manage and bring to the current state personal information of images which are encrypted for safe-keeping clients right not to be public, an of lower quality knowledge-base manager may only get way in to the knowledge for computers putting out of the way key and have to get control of knowledge for computers in encrypted lands ruled over. The order of facts extraction before image decryption gives support to (a statement) the able to be done of our work in this example. When the knowledge-base manager gets the facts putting out of the way key, he can decrypt the LSBplanes of AE and clear substance the added knowledge for computers m by directly reading the decrypted account. When requesting for changing knowledge information of encrypted images, the knowledge-base manager, then, changes knowledge information through LSB putting in place of and encrypts changed knowledge information according to the knowledge for computers putting out of the way key again. As the complete work process is entirely operated on encrypted lands ruled over, it keeps out of the loss of first form what is in. 2) Case 2: Getting from facts FromDecrypted images: In Case 1, both getting fixed and extraction of the facts are made use of, did something with in encrypted lands ruled over. On the other hand, there is a different place, position that the user wants to decrypt the image first and copies from the facts from the decrypted image when it is needed. The supporters example is an attention to for such scenario. take on Alice outsourced her images to a cloud computer, and the images are encrypted to keep safe (out of danger) their what is in. into the encrypted images, the cloud computer marks the images by getting fixed some sign system, including the making-out of the images owner, the making-out of the cloud computer and time stamps, to manage the encrypted images. Note that the cloud computer has no right to do any fixed damage to the images. Now a given authority user, Bob who has been shared the encryption key and the facts putting out of the way key, downloaded and decrypted the images. Bob hoped to get marked decrypted images, i.e., decrypted images still including the system of naming, which can be used to a bit the starting point and history of the facts. The order of image decryption before/without facts extraction is errorlessly right for this example. next, we make, be moving in how to produce a marked decrypted image.

K.S.Parimala, IJRIT

379

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 10, October 2014, Pg. 372-380

4 Conclusion Reversible facts putting out of the way in encrypted images is a new thing talked of picture attention because of the privacy-preserving requirements from cloud facts managers of a business. earlier methods instrument RDH in encrypted images by vacating room after encryption, as opposite to which we made an offer by keeping back roombefore encryption. Thus the facts hider can help from the in addition space put out in earlier stage tomake facts skin, leather process effortless. The made an offer way can take better chances of all old and wise RDH techniques for not beautiful images and get done very good, of highest quality operation without loss of errorless secrecy. In addition, this fiction story way can get done true reversibility, separate facts extraction and greatly getting better on the quality of marked decrypted images.

REFERENCES [1] T. Kalker and F.M.Willems, “Capacity bounds and code constructions for reversible data-hiding,” in Proc. 14th Int. Conf. Digital Signal Processing (DSP2002), 2002, pp. 71–76. [2] W. Zhang, B. Chen, and N. Yu, “Capacity-approaching codes for reversible data hiding,” in Proc 13th Information Hiding (IH’2011), LNCS 6958, 2011, pp. 255–269, Springer-Verlag. [3] W. Zhang, B. Chen, and N. Yu, “Improving various reversible data hiding schemes via optimal codes for binary covers,” IEEE Trans. Image Process., vol. 21, no. 6, pp. 2991–3003, Jun. 2012. [4] J. Fridrich and M. Goljan, “Lossless data embedding for all image formats,” in Proc. SPIE Proc. Photonics West, Electronic Imaging, Security and Watermarking of Multimedia Contents, San Jose, CA, USA, Jan. 2002, vol. 4675, pp. 572–583. [5] J. Tian, “Reversible data embedding using a difference expansion,” IEEE Trans. Circuits Syst. Video Technol., vol. 13, no. 8, pp. 890–896, Aug. 2003. [6] Z. Ni, Y. Shi, N. Ansari, and S. Wei, “Reversible data hiding,” IEEE Trans. Circuits Syst. Video Technol., vol. 16, no. 3, pp. 354–362, Mar. 2006. [7] D.M. Thodi and J. J. Rodriguez, “Expansion embedding techniques for reversible watermarking,” IEEE Trans. Image Process., vol. 16, no. 3, pp. 721–730, Mar. 2007. [8] X. L. Li, B. Yang, and T. Y. Zeng, “Efcient reversible watermarking based on adaptive prediction-error expansion and pixel selection,” IEEE Trans. Image Process., vol. 20, no. 12, pp. 3524–3533, Dec. 2011. [9] P. Tsai, Y. C. Hu, and H. L. Yeh, “Reversible image hiding scheme using predictive coding and histogram shifting,” Signal Process., vol. 89, pp. 1129–1143, 2009. [10] L. Luo et al., “Reversible imagewatermarking using interpolation technique,” IEEE Trans. Inf. Forensics Security, vol. 5, no. 1, pp. 187–193, Mar. 2010.

K.S.Parimala, IJRIT

380