SecretImage SharingBased …dspace.cusat.ac.in/jspui/bitstream/123456789/10260/1...SecretImage SharingBased ChequeTruncationSystem with Cheating Detection Sreela S R a, G Santhosh

Secret Image Sharing Based Cheque Truncation System with

Cheating Detection

Sreela S Ra, G Santhosh Kumara, Binu V Pb

aDepartment of Computer Science, Cochin University of Science and Technology

bDepartment of Computer Applications, Cochin University of Science and Technology

Cheque Truncation System(CTS) is an automatic cheque clearance system implemented by Reserve Bankof India (RBI). CTS uses cheque image, instead of the physical cheque itself, for cheque clearance thusreducing the turn around time drastically. This approach holds back the physical movement of chequefrom presenting bank to the drawee bank. In CTS, digital image of the cheque is protected using standardpublic key and symmetric key encryptions like RSA, triple DES etc.,. This involves a lot of computationoverhead and key management. The security also depends on the hard mathematical problem and is onlycomputationally secure. Information theoretically secure, secret image sharing techniques can be used inthe CTS for the secure and efficient processing of cheque image. In this paper, we propose two simple andefficient secret image sharing schemes and a Cheque Truncation System based on these algorithms. In theproposed scheme, the presenting bank is acting as the dealer and the participants are the customer and thedrawee bank. The dealer should generate the shares of cheque and distributes it to customer and draweebank. The validity of the shares are important during the reconstruction process. The proposed schemealso suggests a method for cheating detection which identify any invalid shares submitted by the customers,using the hashing technique. The experimental results show that the proposed scheme is efficient and securecompared with the existing scheme.

Keywords : Cheque Truncation System, Pixel Expansion, PKI, Secret Image Sharing, Visual Cryp-tography.

1. INTRODUCTION

Cheques represent a significant segment of pay-ment instruments in India. Cheque Trunca-tion System (CTS) or ICS(Image Based Clear-ing System) in India is a project undertakenby Reserve Bank of India ( RBI) for fasterclearing of cheques. CTS is basically an on-line image-based cheque clearing system wherecheque images and Magnetic Ink CharacterRecognition (MICR) data are captured at thecollecting bank branch and transmitted elec-tronically. Manual clearing of cheque needshuman intervention and is a time consumingtask. Cheque truncation [1] involves stoppingthe flow of the physical cheques issued by adrawer to the drawee branch. An electronic im-age of the cheque is sent to the drawee branchalong with the relevant information like the

MICR fields, date of presentation, presentingbanks etc.,.

The point of truncation is left to the discre-tion of the presenting bank. Thus, Chequetruncation, would eliminate the need to movethe physical instruments across branches andhence result in effective reduction in the timerequired for payment of cheques, the associatedcost of transit and delays in processing, etc., .This will speed up the process of collection orrealization of cheques and thus reduce the turnaround time.

The system offers following benefits to thebank and customers. Banks can expect mul-tiple benefits through the implementation ofCTS, like faster clearing cycle,better reconcil-iation/verification process. Besides, it reducesoperational risk by securing the transmission

56

International Journal of Information Processing, 8(4), 56-67, 2014ISSN : 0973-8215IK International Publishing House Pvt. Ltd., New Delhi, India

Secret Image Sharing Based CTS with Cheating Detection 57

route.Reduction of manual tasks leads to re-duction of errors. Customer satisfaction willbe enhanced, due to the reduced turn aroundtime (TAT). Real-time tracking and visibilityof the cheques, less fraudulent cases with se-cured transfer of images to the RBI are otherpossible benefits that banks may derive fromthis solution [2]. For Customers CTS / ICSsubstantially reduces the time taken to clearthe cheques as well increases operational effi-ciency by cutting down on overheads involvedin the physical cheque clearing process. In ad-dition, it also offers better reconciliation andfraud prevention.

The use of the Public Key Infrastructure (PKI)ensures data authenticity, integrity and non-repudiation, adding strength to the entire sys-tem. The presenting bank is required to af-fix digital signature on the images and datafrom the point of truncation itself. The imageand data are secured using the PKI throughout the entire cycle covering capture system,the presenting bank, the clearing house andthe drawee bank. This system needs a lot ofcomputation and overhead in key managementis high. In this paper a secret image sharing[3] based scheme is proposed. Two efficientschemes are proposed which are computation-ally secure and avoids the overhead in key man-agement. A cheating detection scheme is alsoproposed which avoids the use of invalid sharesduring the reconstruction.

In the rest of the paper, Section 2 describesthe CTS Architecture. Section 3 describes therelated work. Proposed system and algorithmsare explained in Section 4. Experimental re-sults are discussed in Section 5 and the conclu-sions are drawn in Section 6.

2. CTS ARCHITECTURE

The process flow of CTS is explained below.In CTS, the presenting bank (or its branch)captures the data on the MICR band and theimages of a cheque using their Capture Systemcomprising of a scanner, core banking or otherapplication. Images and data should meet

the specifications and standards prescribed fordata and images. The architecture of CTS isexplained in Figure 1.

To ensure security, end-to-end Public Key In-frastructure (PKI) has been implemented inCTS for protecting data and image. The pre-senting bank sends the data and captured im-ages duly signed and encrypted to the ClearingHouse (the central processing location) for on-ward transmission to the paying bank (destina-tion or drawee bank). For the purpose of par-ticipation the presenting and drawee banks areprovided with an interface / gateway called theClearing House Interface (CHI) that enablesthem to connect and transmit data and imagesin a secure and safe manner to the ClearingHouse (CH).

The CTS uses Public Key Infrastructure (PKI)like digital signature and encryption for pro-tecting cheque images and data. The stan-dards defined for PKI are hash algorithm SHA-1, padding algorithm, RSA asymmetric en-cryption with 1024 bit key length, Triple DES(3DES, TDES) symmetric encryption with 168bit key length and Certificates in x.509v3 for-mat. Cheque image is protected using encryp-tion techniques. These techniques need a lot ofcomputation and usage of keys.

3. RELATED WORK

CTS system is implemented by RBI to reducethe complexity of cheque processing. CTS sys-tem is implemented in India in 2010. Gridbased CTS is implemented in Chennai, Delhi,Kolkata etc., . The different security schemesare applied in cheque. Pasupathinathan et al.,

[4] describes privacy enhanced electonic chequesystem in 2005. In 2011, Rigel Gjomemo et

al., [5] explains the digital cheque Forgery at-tack on CTS. Kota, Saranya and Rajarshi Pal[6] explains the method for detecting tamperedcheque images in CTS Using Difference Expan-sion Based Watermarking in 2014.

The secret image sharing schemes are based onvisual cryptography, number theory [7], infor-mation hiding theory, error diffusion technique,

58 Sreela S R, et al.,

Figure 1. CTS Architecture

boolean operation etc., . In Yan, Xuehu scheme[8], secret image sharing is based on informa-tion hiding theory. The important techniqueused in this scheme are MLE and LSBM. Butthis scheme is applicable only to the binary im-ages. Chen and Chang [9] use quadratic residuetechnique for secret image sharing. They pro-posed a (2, 2) scheme which is lossy and thelossless scheme having the share size largerthan the secret. The computations involved isalso more. Chen, Wei-Kuei and Hao-Kuan Tso[10] introduced a secret image sharing schemefor protecting medical images using Hill ci-pher method. Thein-Lins Scheme enhancedShamir’s secret sharing scheme [11](Lagrangeinterpolating polynomial) for protecting digitalimages. Table 1 explains a comparative studyon different secret image sharing schemes.

4. PROPOSED SYSTEM

The system architecture describes how secretimage sharing scheme happening in the CTS.In this architecture, the dealer should be thepresenting bank. The participants are cus-tomer, clearing house (CH) and drawee bank.Figure 2 explains the system architecture.

In order to reduce the computation and usageof keys, cheque image can be protected usingsecret image sharing. In this paper, two se-

cret image sharing methods are proposed forprotecting cheque images. If any one of theparticipants do malpractice on the shares, thencheating occurs. Cheating detection is imple-mented in this paper.

In secret image sharing technique, a secret im-age is distributed to some of the participantsthrough splitting the image into different piecescalled shares and recover the secret image bycollecting the sufficient number of shares fromauthorized participants. This field of cryptog-raphy is called visual cryptography or visual se-cret sharing [12]. If any one of the participantdo malpractice on their shares, cheating detec-tion methods can be used. It consists of threephases: share generation phase, distributionphase and reconstruction phase. In the sharegeneration phase, the digital image is split intodifferent pieces called shares. In the distribu-tion phase, the shares are distributed to au-thorized participants and in the last phase, theimage is reconstructed using sufficient numberof shares from authorized participants.In a se-cret image sharing scheme, there is a secret im-age S to be shared among a set of participants.The secret is known to a special person calleddealer. The dealer generates and distributespartial information called shares to the partic-ipants.


Table 1Comparison of Different Schemes

Scheme (k,n)threshold

RecoveringMeasure

Loss-less

PixelExpansion

size ofshare

VCS [12] (k,n) Stacking No Yes IncreasesExtendedVCS[13]

(k,n) Stacking No Yes Increases

(k, n2) [8] (k, n2) Mod andBoolean /addition andcomparison

Yes No sameasoriginalimage

(2,3) [7] (2,3) ModandMultiplication

Yes No sameasoriginalimage

Thein-Lin

[14] (r , n) ShamirsSSS(LagrangeInterpolatingpolynomial)

No No Reducesbyhalf

BooleanVSS

[15] (r , n) Booleanoperations

Yes No Increases

(2,3) scheme is required for implementing se-curity in CTS. Presenting bank generates theshares and distributed to the clearing house,drawee bank and to the customer. Customershould use the share to get the information ofprocessing of cheque through online. Draweebank should reconstruct the cheque image us-ing the share from the CH and its own share.Drawee bank cant reconstruct the cheque im-age using its own share. To implement securityin CTS, xor scheme and partition scheme canbe used.

The important steps involved in the proposedCTS using secret image sharing are as follows:

1. Customer submits the cheque to the pre-senting bank.

2. Capture image of cheque and data usingcapture system

3. Send the data and image to the present-ing CHI.

4. Presenting CHI provide security to thecheque image using (3,2) secret imagesharing scheme.

5. Send first share of the cheque image(SC1)and data to the clearing house throughthe CHI.

6. Send second share of the cheque image(SC2) to the customer if customer sub-mits cheque through online and this shareis used for authentication for viewing thedetails of cheque processing.

7. The clearing House send data and oneshare of the cheque image(SC1) to thedrawee bank through receiving CHI.

8. The drawee bank request another shareof the cheque image from the presentingbank through receiving CHI.

9. The presenting bank submit third share(SC3) to the drawee bank.


Figure 2. System Architecture

Figure 3. System Architecture

10. The receiving CHI reconstructs thecheque image using shares SC1 and SC3.

11. Send data to the drawee bank for pro-cessing cheque.

12. Bank process the cheque using image pro-cessing algorithm.

In Figure 3, the numbers represent the abovesteps.

4.1. XOR scheme

XOR scheme is a (2,3) scheme. In this scheme,three shares are created and the original im-age is reconstructed using at least two shares.The image is not reconstructed using only oneshare. The share images are created by divid-ing pixel into four bits. In this scheme, the

share image pixel is 4 bits. Share generationalgorithm is explained in Algorithm 1. Recov-ery algorithm is explained in Algorithm 2. Theoriginal secret image is reconstucted by usingany of the two shares from three shares.

Consider an image matrix is

157 160 190 13089 255 224 19210 220 255 22464 128 192 255

Consider the secret image pixel is 190. Its bi-nary representation is 10111110. The share1pixel(SC1=6(0110)) is created using even bits.The share2 pixel(SC2=15 (1111)) is created us-ing odd bits. The share3 pixel(SC3=9(1001))is created by XOR of s1 s2. The XOR schemeis applied on the above image S. The threeshares obtained SC1, SC2 and SC3 are as fol-lows:SC1=

7 0 6 013 15 8 80 14 15 88 0 8 15


SC2=

10 12 15 92 15 12 83 10 15 120 8 8 15

SC3=

13 12 9 915 0 4 03 4 0 48 8 0 0

In this scheme, the size of the share is half of

Algorithm 1: Algorithm: Share Generation1 Input: M X N Secret grayscale image S2 Output: Share images SC1, SC2, SC33 begin

1. For each pixel (i, j)ε{(i, j)|1 ≤ i ≤ M, 1 ≤ j ≤ N}repeat steps 2- 4

2. Pixelvalue, pv= S(i,j) which is the binary arraycontaining the pixel intensity binaryrepresentation.

3. Create share1 SC1(i, j) pixel using even bits ofS(i,j) pixel

SC1(i, j) =

3∑

k=0

(pv(2k) × 2k)

4. Create share2 SC2(i,j) pixel using odd bits ofS(i,j) pixel

SC2(i, j) =

3∑

k=0

(pv(2k + 1) × 2k)

5. Create share3 SC3(i,j) pixel by xor ing SC1(i, j)and SC2(i, j)

SC3(i, j, k) = SC1(i, j) ⊕ SC2(i, j)

6. Output shares SC1, SC2, and SC3.

end

the size of the original image. The number ofbits for representing a pixel in share is 4 bits.If the M ×N secret gray scale image has a sizeof 8×M ×N bits, then the size of the share isonly 4 ×M × N bits. So the storage space ofthe share is reduced. The quality of the recon-structed image is same as the original image.In this scheme, there is no pixel expansion onreconstructed image. It is a lossless scheme.

Algorithm 2: Recovery Algorithm1 Input: Share images SC1, SC2, SC32 Output: Reconstructed Secret Image S3 begin4 The secret image can be reconstructed from shadow

images SC1, SC2

1. For each position,(i, j)ε{(i, j)|1 ≤ i ≤ M, 1 ≤ j ≤ N} repeat step2.

2. S( i , j) is obtained by intermixing bits of SC1(i, j)and SC2(i, j) in even and odd positionsrespectively.

3. Output image S.

The secret image can be reconstructed from shadowimages SC1, and SC3

1. For eachposition,(i, j)ε{(i, j)|1 ≤ i ≤ M, 1 ≤ j ≤ N}repeat step 2-3.

2. b = SC1(i, j) ⊕ SC3(i, j)

3. S(i, j) is obtained by intermixing bits of SC1(i,j)and b in even and odd positions respectively.

4. Output image S.

The secret image can be reconstructed from shadowimages SC2, and SC3.

1. For each position,(i, j)ε{(i, j)|1 ≤ i ≤ M, 1 ≤ j ≤ N} repeat step2-3

2. b = SC2(i, j) ⊕ SC3(i, j)

3. S(i, j) is obtained by intermixing bits of b andSC2(i,j) in even and odd positions respectively.

4. Output image S.

end

4.2. Partition Scheme

Partition scheme is a (2,3) scheme. It usesboolean XOR operations. This method usesrandom number for creating shares. The sharegeneration algorithm is explained in Algorithm3. Algorithms 4, 5, 6 describe the reconstruc-tion of image.

4.3. Cheating Detection Scheme using

Hash Function

A threshold scheme for secret sharing can pro-tect a secret with high reliability and flexibil-ity. These advantages can be achieved onlywhen all the participants are honest, i.e., allthe participants willing to pool their shadowsshall always present the true ones. Cheating


Algorithm 3: Algorithm: Share Generation1 Input: M X N Secret grayscale image S2 Output: Share images SC1, SC2, SC33 begin

1. Let s be the pixel of the secret image(S) and r bea random number in 0-255.

2. s is divided into s1 and s2 and r into r1 and r2.

3. Share1 pixel is created by combining s2 ⊕ r2 andr1

4. Share2 pixel is created by combining s1 ⊕ r1 andr2

5. Share3 pixel is created by combining s2 ⊕ r1 ands1 ⊕ r2

6. repeat step 1-5 until all pixels of image areprocessed.

7. Output three shares share1(SC1), share2 (SC2),share3(SC3).

detection is an important issue in the secretsharing scheme. However, cheater identifica-tion is more effective than cheating detection inrealistic applications. If some dishonest partic-ipants exist, the other honest participants willobtain a false secret, while the cheaters mayindividually obtain the true one. By apply-ing a one-way hashing function along with theuse of arithmetic coding, the proposed methodcan be used to deterministically detect cheat-ing and identify the cheaters, no matter howmany cheaters are involved in the secret recon-struction.

Two important theorems used in cheating de-tection using hash function are as follows. Letai be the random shares of the secret data andp be the randomly generated prime number.Theorem 1 [16]: Let T =

∑n

i=1 aipi−1, where

0 ≤ ai < p. Then

⌊T

pj−1⌋(mod p) = aj (1)

Extended from Theorem 1, we have the follow-ing result.Theorem 2 [16]: Let T =

∑n

i=1 aip2(i−1)) +

∑n−1i=1 cp2i−1, where −p < ai < p and 1 ≤ c <

Algorithm 4: Algorithm: Reconstruction us-ing Share1 and Share2

1 Input: Share images SC1, SC22 Output: Reconstructed Secret image S3 begin4 Original image is reconstructed from share1 and share2

by applying following steps.

1. The share1(sc1) pixel is divided into two equalparts sc11 and sc12.

2. The share2(sc2) pixel is divided into two equalparts sc21 and sc22.

3. The second part of the original image pixel (s2) isreconstructed by XOR-ing first part of the share1pixel(sc11) and second part of the share2pixel(sc22).

s2 = sc11 ⊕ sc22

4. The first part of the original image pixel(s1) isreconstructed by XOR-ing second part of the theshare1 pixel(sc12) and first part of the share2pixel(sc21).

s1 = sc12 ⊕ sc21

5. The original image pixel(s) is obtained bycombining s1 and s2.

s = s1.s2

6. repeat the above steps until all pixels areprocessed.

7. Output image S

end

p. Then

⌊T

p2(j−1)⌋(mod p) = aj(mod p) (2)

Combining this result with secret image shar-ing scheme, the following method is used forcheating detection and cheater identification.Algorithm for cheating detection and cheateridentification is explained in Algorithm 7.

The hash value of the image is generated usingcontent of the image. The hash value of the im-age is also generated using feature vector of theimage. In the secret image sharing, any simplechange in the shares is treated as a cheating.Any mild change in the image is reflected inthe hash value of image using content of im-age rather than using feature vector of image.



1 Input: Share images SC1, SC32 Output: Reconstructed Secret Image S3 begin4 Original image is reconstructed from share1 and share3


1. The share1 pixel is divided into two equal partssc11 and sc12.


3. The second part of the original image pixel(s2) isobtained by XOR-ing second part of the share1pixel(sc12) and first part of the share3 pixel(sc31).

s2 = sc12 ⊕ sc31

4. b = s1 ⊕ s2 is obtained by XOR-ing first part ofthe share1 pixel(sc11) and second part of theshare3 pixel(sc32).

b = sc11 ⊕ sc32

5. The first part of the original image pixel(s1) isobtained by s1 = b ⊕ s2

6. Secret image pixel(s) is reconstructed bycombining s1 and s2

s = s1.s2

7. Repeat above steps until all pixels are processed.

8. Output image S

end

So we use the hash generation method usingcontent of the image.

4.4. Cheque Processing

Cheque processing is implemented in Draweebank. In our work, the courtesy amount re-gion and account number field is recognized.The important steps associated with chequeprocessing are as follows:

1. Load cheque in grayscale.

2. Find courtesy amount region in chequeusing cheque template method.

3. Find account number region in chequeusing cheque template method.

4. Segment digits in courtesy amount andresize each digit having a size of 28× 28.


1 Input: Share images SC2, SC32 Output: Reconstructed Secret image S3 begin4 Original image is reconstructed from share2 and share3




3. The first part of the original image pixel(s1) isobtained by XOR-ing second part of the share2pixel(sc22) and second part of the share3pixel(sc32).

s1 = sc22 ⊕ sc32

4. b = s1 ⊕ s2 is obtained by XOR-ing first part ofthe share2 pixel(sc21) and first part of the share3pixel(sc31).

b = sc21 ⊕ sc31

5. The second part of the original image is obtainedby s2 = b ⊕ s1

6. Secret image pixel(s) is reconstructed bycombining s1 and s2.

s = s1.s2

7. Repeat above steps until all pixels are processed.

8. Output image S

end

5. Apply digit recognition method for rec-ognizing digit in courtesy amount.

6. Combine each digit and generate cour-tesy amount.

7. Segment digits in account region and re-size each digit having a size of 28× 28.

8. Apply digit recognition method for rec-ognizing digit in account number.

9. Combine each digit and generate accountnumber.

10. Process the amount from the account anddeduct the amount from the account.

11. Send the information to the presentingbank through clearing house.


Algorithm 7: Cheating Detection andCheater Identification using Hash Function

1 Dealer generates the shares for cheque image using secretimage sharing algorithm.

2 He generates public parameters T and p in the followingsteps.

3 Choose a one-way function h(.) and a prime number p

such that h(.) < p. Generates hash value of image usinghash function.

4 Compute T =∑n

i=1 h(si)p(2(i−1)) +

∑n−1i=1 cp2i−1 where

c is a positive constant randomly chosen over GF (p)5 Publish T and p.6 Dealer distributes shadow SCi to participants Ui. for

i = 1, 2, ..., n.7 In the receiver side, cheating detection and cheater

identification can easily be achieved by applying thefollowing procedure.

8 Participants UjǫG present their possessed shadows SC′

j

and compute T ′ =∑

UjǫGh(SC′

j)p(2(i−1))

9 For each UjǫG, check ⌊ T−T ′

p(2(j−1))⌋(mod p)

?= 0

10 If the equation holds, participantUj is honest; otherwise,Uj is a cheater.

12. At last customer gets the amount fromthe presenting bank.

4.4.1. Digit Recognizer

In our work, digit recognition is done using K

Nearest neighbour classification technique [17].The isolated components after segmentationare fed into a digit recognizer. The accuracy inrecognizing constituent digits plays a big rolein the recognition accuracy of the handwrittencourtesy amount numeral string [18-30]. Af-ter successful segmentation of individual digitsfrom the numeral string, they have to be cor-rectly recognized to get the value of the cheque.In this, there is two steps: training phase andtesting phase. In training phase, hand writtenimages are trained. In the testing phase, thefollowing steps need to be carried out.

• The digit in the image is centered.

• Convert two dimensional array to one di-mensional array using reshape operation.

• Apply the one dimensional array to theKNN classifier.

• The digit is recognized as output.

5. EXPERIMENTAL RESULTS

The algorithms are implemented in Java.The experimental result obtained for partition

scheme using the 500 × 225 gray scale chequeimage is shown in Figure 4. The reconstructedimage has the same quality as original image.This algorithm is also useful for color images. Ifthis algorithm is applied in color images, the al-gorithm is applied on each channel (Red, Blue,Green) separately. The bit depth of the shareof color image is 12 bits. So this scheme is en-hanced for color images also. The comparisonof above schemes are described in Table 2.

(a) Secret Image

(b) Share1 (c) Share2 (d) Share3

(e) Recon-structed Image

Figure 4. Result of Partition Scheme

The mean square error(MSE) is used to mea-sure the mean square error between original(I)and recovered image(I’) and is calculated byusing the equation

MSE =1

MN

∑

i=1,M

∑

j=1,N

(I ′(i, j)− I(i, j))2

The MSE between original and recovered im-age is 0.

In the cheating detection phase, the hash valueof the share images are calculated in the senderside. The value of T is 4.59080713E8. In thereceiver side, the hash value is not computed.The value of T’ is computed in the receiver side.If the remainder is zero, the cheating does notoccur in the shares of the cheque image. Ifthe cheating does not occur in the shares, thecheque image is reconstructed from the shares.Otherwise, the drawee bank request for the cor-rect shares from the participants.


Table 2Property Comparison of Proposed Schemes

Scheme RecoveringMeasure

Loss-less

PixelExpansion

size ofshare

XOR Boolean Yes No half oftheimagesize

Partition Boolean Yes No sameaschequeimage

In cheque processing, the courtesy amountand account number region are recognised us-ing image processing technique. The courtesyamount region in cheque image is shown in Fig-ure 5

Figure 5. Courtes Amount

The account number region in cheque imageis shown in Figure 6 Each digit in courtesy

Figure 6. Account Number

amount and account number are segmentedand applied to the digit recognizer. For digitrecognition using KNN, the standard datasetMNIST handwritten digit image is used astraining set. MNIST dataset contains 60000image for training purpose. The KNN clas-sifier give correct result for MNIST testingimages. Some misclassification occured forcourtesy amount recognition. The courtesyamount recognised in Figure 5 is 125795.75.The account number reconised in Figure 6is 12781151507. The account number andamount is fed to the core banking softwareand do the transaction operations in software.Drawee bank returns the transaction details or

error message to the presenting bank throughClearing House.

6. CONCLUSIONS

Cheque Truncation system accelerates the pro-cess of collection of cheques resulting in bet-ter service to customers, reduces the scopefor clearing-related frauds or loss of instru-ments in transit, lowers the cost of collection ofcheques and removes reconciliation-related andlogistics-related problems, thus benefit the sys-tem as a whole. In this paper, two secret imagesharing schemes are proposed for provid- ingsecurity to the cheque image in the CTS. Theproposed XOR scheme is simple and effecientbut it is not ideal. It can be used in low stor-age device where memory is a contsraint. Theshare size is only half of the original image andit is a lossless scheme. Partition scheme havethe properties such as no pixel expansion andlossless scheme. The scheme is also ideal.

The experimental result shows that the pro-posed system provides better security and ef-ficiency in Cheque Truncation System (CTS).The operations invloved are simple XOR andit also avoids the complicated encryption de-cryption operations which are time consuming.The secret image sharing scheme doesn’t needany key management and authentication of theshares are done with simple hash function. Theshares are also verified with the help of publicparameters. We are looking forward for im-


proved cheque processing using advanced im-age processing technique which helps in auto-matic cheque processing. The operational effe-ciency, speed accuracy, security and authenti-cation are the major design objectives.

REFERENCES

1. RBI. RBI CTS. http://www.rbi.org.in.2. Cheque Truncation System. http://www.

banknetindia.com/banking/chqtruncation.htm.3. Sreekumar A. Secret Sharing Schemes using

Visual Cryptography, In Dissertation, CochinUniversity of Science and Technology, 2009.

4. Pasupathinathan, Vijayakrishnan, JosefPieprzyk and Huaxiong Wang. Privacy En-hanced Electronic Cheque System, CEC.2005.

5. Rigel Gjomemo, Ha z Malik, Nilesh Sumb, VN Venkatakrishnan and Rashid Ansari. DigitalCheck Forgery Attacks on Client Check Trun-cation Systems, Signal Processing 91.1, pages90–97,2011.

6. Kota, Saranya and Rajarshi Pal. DetectingTampered Cheque Images in Cheque Trunca-tion System using Difference Expansion BasedWatermarking, IEEE International AdvanceComputing Conference (IACC), IEEE, 2014.

7. Binu V P and A Sreekumar. Lossless SecretImage Sharing Schemes, International Journalof Computational Intelligence and InformationSecurity, 2013.

8. Xuehu Yan, Shen Wang, Ahmed A Abd El-Latif and Xiamu Niu. New Approaches forEfficient Information Hiding-based Secret Im-age Sharing Schemes, Signal, Image and VideoProcessing, pages 1–12, 2013.

9. Chang-Chu Chen and Chin-Chen Chang. Se-cret Image Sharing using Quadratic Residues,Intelligent Information Hiding and MultimediaSignal Processing, 1:515–518, IEEE, 2007.

10. Wei-Kuei Chen and Hao-Kuan Tso. VisualSharing Protection Method for Medical Im-ages, Journal of Medical Systems, 37(1):1–8,2013.

11. Adi Shamir. How to Share a Secret, Commu-nications of the ACM, 22(11):612–613, 1979.

12. Adi Shamir Naor, Moni. Visual Cryptogra-phy, Advances in CryptologyEUROCRYPT’94,pages 1–12, 1995.

13. Ateniese, Giuseppe, et al. Extended Capabili-ties for Visual Cryptography, Theoretical Com-puter Science, pages 143–161, 2001.

14. Thien, Chih-Ching and Ja-Chen Lin. SecretImage Sharing, Computers and Graphics 26.5, pages 765–770, 2002.

15. Chen, Tzung-Her and Chang-Sian Wu. Effi-cient Multi-Secret Image Sharing Based onBoolean Operations, Signal Processing 91.1,pages 90–97, 2011.

16. Wu T C and T S Wu. Cheating Detectionand Cheater Identification in Secret SharingSchemes, IEEE Proceedings on Computers andDigital Techniques, 142(5): IET, 1995.

17. Larose, Daniel T. kNearest Neighbor Algo-rithm, Discovering Knowledge in Data: An In-troduction to Data Mining, pages 90–106, 2005.

18. CTS Clearing House Interface Specification,NCR Corporation and RBI/NPCI, 24th Au-gust, 2010.

19. Calin A Sandru. 125 Neptune Drive, Suite 606,North York (CA), M6A 1X3 Apparatus andMethod for Enhancing the Security of Nego-tiable Documents, Patent No.: US 6, 233, 340B1, Date of Patent: May 15, 2001.

20. R Jayadevan, S R Kolhe, P M Patil, U Pal.Automatic Processing of Handwritten BankCheque Images: A Survey, International Jour-nal on Document Analysis and Recognition(IJDAR) 15(4):267–296, 2012.

21. Chen, Yu-Chi, Du-Shiau Tsai and GwoboaHorng. Visual Secret Sharing with CheatingPrevention Revisited, Digital Signal Process-ing, 23(5):1496–1504, 2013.

22. Chen, Yu-Chi, Du-Shiau Tsai and GwoboaHorng. ”A New Authentication Based Cheat-ing Prevention Scheme in NaorShamirs VisualCryptography, Journal of Visual Communi-cation and Image Representation, 23(8):1225–1233, 2012.

23. Zhao, Rong, et al. A New Image Secret SharingScheme to Identify Cheaters, Computer Stan-dards and Interfaces, 31(1):252–257, 2009.

24. Ahmed, Fawad, Mohammed Yakoob Siyal andVali Uddin Abbas. A Secure and Robust Hash-based Scheme for Image Authentication, SignalProcessing, 90(5):1456–1470, 2010.

25. Askari, Nazanin, Cecilia Moloney and HowardHeys. Application of Visual Cryptography toBiometric Authentication, 2011.

26. Fatma, E-ZA Elgamal, Noha A Hikal and F EZ Abou-Chadi. Secure Medical Images Sharingover Cloud Computing Environment, Interna-tional Journal ON Cloud Computing, 2013.

27. Guo, Huiping and Nicolas D Georganas. A


Novel Approach to Digital Image Watermark-ing Based on a Generalized Secret Shar-ing Scheme, Multimedia Systems 9(3):249–260,2003.

28. Horng, Gwoboa, Tzungher Chen and Du-ShiauTsai. Cheating in Visual Cryptography, De-signs, Codes and Cryptography, 38(2):219–236,2006.

29. Lee, Yuchun. Handwritten Digit Recognitionusing k Nearest-Neighbor, Radial-Basis Func-tion and Backpropagation Neural Networks,Neural computation, 3(3):440–449, 1991.

30. Guillevic, Didier and Ching Y Suen. HMM-KNN Word Recognition Engine for BankCheque Processing, in Proceedings of Four-teenth IEEE International Conference onPattern Recognition, 2, 1998.

S R Sreela is a ResearchScholar in the Departmentof Computer Science, CochinUniversity of Science and Tech-nology(CUSAT). She holds aBachelor Degree in InformationTechnology and Masters Degreein Computer and InformationScience. Her research area inclu-

des Image Processing, Secret Sharing and Security.

G Santhosh Kumar receivedhis MTech Degree in Computerand Information Science fromCUSAT, in 1999 and Ph.D inWireless Sensor Network fromCochin University of Science andTechnology. Currently he is wo-

rking as an Associate Professor in CUSAT. He hadmore than 15 years of teaching experience. His re-search interest includes Wireless Networks, MobileCommunications and Software Architecture.

V P Binu is a ResearchScholar in the Department ofComputer Applications, CochinUniversity of Science and Tech-nology(CUSAT). He holds aBachelor Degree in ComputerScience and Engineering andMasters Degree in Computer

and Information Science. His research area in-cludes cryptography, secret sharing and security.