World Applied Programming, Vol (2), Issue (1), January 2012.

42-48
Special section for proceeding of International e-Conference on Computer Engineering (IeCCE) 2012

ISSN: 2222-2510
2011 WAP journal. www.waprogramming.com

An Efficient Secret Image Sharing Scheme

Vinay Rishiwal1

Ashutosh Gupta * 2

CSIT Department, MJP Rohilkhand University

Bareilly, India
1
vinayrishiwal@gmail.com

CSIT Department, MJP Rohilkhand University

Bareilly, India
2
ashutosh333@rediffmail.com

Abstract: Secret Sharing is one of the efficient methods for secured transmission of data. This paper discusses
the flaws associated with the existing scheme of secret sharing for image transmission. An efficient secret
sharing scheme as well reconstruction scheme has also been proposed which overcomes flaws in existing
scheme. It has been shown through experimental results that the proposed scheme can securely transmit the
image of any size and can effectively recover the same at the receiver end.
Keywords:

Secret Sharing, Image Transmission.

I.

INTRODUCTION

Secret transmission of data is an important task to preserve the data from the probable threats, during the
transmission. Various techniques [1,2,3,4,5,6,11] have been proposed in literature for secure transmission of data
but not much work has been done on the secret transmission of images, which is a very difficult task to
accomplish. One of the secret sharing schemes [11,7] which are used in literature to share the image at
transmission side can be applied either by using Blakeley's secret sharing scheme [2]) or Shamir's scheme [1] for
sharing a secret. In this paper geometry based Blakeley's scheme is analyzed. The motivation for secret sharing
[1,2] comes from the concept of secure key management. In some situations, there is usually one secret key that
provides access to many important files. If such a key is lost (e.g., the person who knows the key becomes
unavailable, or the entire computer which stores the key is destroyed), then all the important files becomes
inaccessible. The basic idea in secret sharing is to divide the secret key into pieces and distribute the pieces to
different persons so that certain subsets of the persons can get together to recover the key. The secret image
sharing approach has been introduced by Adi Shamir [1].
Later, the secret data can be reconstructed if there is at least completely knowledge of k shadows, where k n.
In 1995, Noar and Shamir first extended the application of this concept to the binary images domain [22], named
visual secret sharing (VSS). Noar and Shamir's VSS scheme eliminates the complex computation problem, and
the secret image can be restored by stacking operation. In addition, the reconstructed image can be just identified
by the human visual system. Over the past decade, most VSS schemes for images have been implemented by
simply stacking the collected shadows. Without computational complexity, but pixel expansion and low
accuracy are weaknesses of these schemes [13,17,19,21,22]. Later, other VSS schemes have been proposed to
resolve these above deficiencies, but they have suffered from the problem of high computational complexity
[14,18,23]. To deal with these problems, Ito et al. [20], Wang et al. [24], and Yang et al. [25] applied the
probability concepts to design a probabilistic visual secret sharing scheme for binary images, called (2, n)
ProbVSS. However, all these existed (2, n) ProbVSS schemes have been designed for binary images and have
achieved low reconstructed secret image quality. In 2008, Chang et al. [15,16] extended the application of
ProbVSS to encompass both grayscale images and color images, respectively. The experimental results show
that Chang et al.'s (2, n) ProbVSS schemes are superior to those existing ProbVSS schemes with respect to four
criteria.
Harn and Lin [8] have proposed a general secret sharing policy to solve the secret/shadow regeneration problem
by extending the life span of the shadows. A generalized secret sharing scheme is proposed by Chang and Tsai

42

Ashutosh Gupta et al., World Applied Programming, Vol (2), No (1), January 2012.

[9]. An efficient secret sharing schemes as well as reconfiguration scheme has been discussed in [10].An
approach to share binary secret images by piling the binary image has been discussed in [7]. Blundo et al. [3]
have proposed a secret image sharing approach for grey level images. Thein and Lin [4] have described a
numerical processing based approach to share a protected image secretly in which they first permute an image
into random image and then shared images are generated using secret sharing approach proposed by Shamir [1].
In [11] Brinkman et al. have proposed a secret sharing scheme to search the contained image in encrypted data. It
is also shown that to securely recover the image at destination any of the secret sharing scheme proposed by
Blakley [2] and Shamir [1] can be used. A hierarchical secret image sharing method has been discussed by
Hastimoto et al. [12] where usability of image information has been increased by using JPEG 2000 code stream
syntax. In Blakley (k,n) secret sharing scheme [2], a secret is divided among n players, such that any k number
of player are needed to reconstruct the image at receiver side and no secret is revealed with the shares less then k
players .
This paper first analyzes the Blakley's secret sharing scheme and then discusses flaws in this approach. It also
presents efficient schemes which overcome all the flaws. The scheme is based on image segmentation
techniques. The proposed scheme combines a subset images to recover the image at the receiver end.
The paper is organized as follows. Section 2 presents a review of Blakley's secret sharing scheme. The flaws in
Blakley's secret sharing scheme are discussed in Section 3. The proposed algorithms are presented in Section 4.
Section 5 discusses the experimental results. Conclusion is given in section 6.
II.

REVIEW OF BLAKLEY'S SECRET SHARING SCHEME

Blakley's secret sharing scheme [2] is geometric in nature. The secret is a point in an m - dimensional space and
n shares are constructed in such a way that each share defines a hyper-plane in this space. The secret can be
revealed by finding the intersection of any m of n planes. This scheme can be modified to achieve perfect
security where two shares are required to recover the secret. A 2 - dimensional plane is used as only two shares
are required to recover the secret. Each share is a line that passes through the point of intersection. If any two of
the shares are put together then the secret which is the point of intersection, can be easily derived. The secret
message is a point in a k - dimensional space and n shares are affine-hyper planes that intersect at this point. The
solution set x =(x1, x2, , xk) to an equation a1x1+a2x2 + akxk = b forms an affine hyper-plane.
Suppose there is a secret image to be shared among n participants. The picture is divided into n transparencies
(shares) such that if any m transparencies are placed together, the image becomes visible but if fewer than m
transparencies are placed together nothing can be seen. Such a scheme is constructed by viewing the secret
image as a set of black and white pixels and handling each pixel separately. The schemes in perfectly secure and
can be implemented easily.
III.

FLAWS IN EXISTING SECRET SHARING ALGORITHMS

Blakley's secret sharing scheme has been applied to images by Thein et al. [4]. The scheme given in [2] has some
short comings (flaws) which can play an important role in image reconstruction and may lead to faulty
reconstructed images at received end.
A. Blakley's Sharing Algorithm
At the sender end, sharing process undergo the following steps from chosen number of shares k to retrieve the
secret with total number of shares n.
Step 1: Divide the original image in set of k pixels to represent the x =(x1, x2, , xk) in the given plane.
Step 2: Choose n sets of (a1, a2, ,ak, b) such that equation $ a1x1+a2x2 + akxk = b is satisfied, where (a1, a2, ,ak,
b) are the parameters.
Step 3: Repeat steps 1 and 2 for every set of k pixels and get n equations.
Step 4: From the k shares by putting the values got from the solution of above equations.

43

Ashutosh Gupta et al., World Applied Programming, Vol (2), No (1), January 2012.

The above presented secret image sharing scheme has two flaws which affect the efficiency of the algorithm are
discussed below.
Flaw 1: The total number pixels in the image should be an integer multiple of k to partition the image into non
overlapping set of k pixels. It is observed during the implementation that the above discussed scheme failed to
recover the image correctly if n is not an integer multiple of k.
Flaw 2: In the above discussed image sharing scheme, the size of parameters (a1, a2, ,ak, b) is determined
mathematically and each parameter requires k * 8/(k + 1) bits. This sometimes creates a difficulty if number of
bits is not an integer which can happen quite often. For example if k =5 then number of bits comes to be 5 *
8/(5+1)=6.66.
The above mentioned flaws have been taken care in this work and proposed scheme along with the possible
solutions to these flaws are discussed in Section 4.1.1.

B. Blakley's Recovery Algorithm

The encrypted transmitted images one reconstructed at the receiver end by solving the equations formed by
considering k} shares from total n} parameters. The reconstruction process is as follows:
Step 1: For preset values of k and n use the same threshold (k,n).
Step 2: Each shared image is partitioned into no overlapping sets of k pixels.
Step 3: Acquire k +1 parameters (a1, a2, ,ak, b) from k pixels.
Step 4: Find the intersection point of these k hyper-planes constructed by parameters.
Step 5: Store the coordinate of the intersection point in a k - dimensional space as k pixels in the reconstructed
image.
Step 6: Repeat Step 2 through Step 5 to reconstruct the transmitted image.
The scheme given in [4] cannot be generalized due to the flaws discussed in Section 3.1. Therefore there is a
need to develop a new generalized scheme. In the next section we propose a generalized scheme which works
well on variety of images.
IV.

PROPOSED IMAGE SHARING SCHEME

The first flaw mentioned in Section 3.1 is a major problem. This problem can be solved in following three ways:
1.

2.
3.

When the image is divided in pixels and pixel assets are formed by combining groups of pixels, the
brightness value of the pixels left alone are treated as it is without any mathematical computation. This
does not create any problem in security as by sending one or two such values, secret cannot be retrieved by
other party.
Overlap the remaining pixels with some of the pixels of previous group of k pixels to form a new group of
k pixels.
Make one of the set of k pixels large than k to cover all the remaining pixels.

The solution of flaw 2 must be such that decimal values should not appear in the calculation of bits required for
the parameters (a1, a2, ,ak, b). This may be solved by using k images for a (k,n) threshold scheme. The earlier
image sharing scheme, uses only a single image to be shared, transmitted and reconstructed. When this single
image is divided in set of k pixels, the grey scale values of many adjacent pixels are sometimes approximately
same. Hence, many problems do arise at the time of parameter computation. In a particular set of pixels, the
proposed scheme in place of a single image uses gray scale values of more than one image to get rid of problem
of similar gray scale values in the neighborhood. The scheme is described in detail in the following section.

44

Ashutosh Gupta et al., World Applied Programming, Vol (2), No (1), January 2012.

A. Image Sharing Algorithm

The proposed algorithm uses three images instead of single image. These three images are transmitted
and shared using the steps given in algorithm IMAGE_SHARE.
Algorithm: IMAGE_SHARE
Step 1: Divide each image into sets of k pixels.
Step 2: Take first pixel of each image and form a group of pixels.
Step 3: The grayscale value of these pixels set are determined and represented in a matrix.
Step 4: Form the equation (a1, a2, ,ak, b).
Step 5: Using the equation of Step 4 compute the values of parameters for first pixel set and give to first shared
image.
Step 6: Similarly, the values of parameters of next pixel set are computed and given to second and third image
then again to first shared image.
Step 7: Repeat Step 6 till all the pixel sets of the three images are considered.
The above algorithm encrypts the input image using two or more random images so that the intruders do not get
the exact information about the original image. In the next section we describe the algorithm for reconstruction
of input image at recover end.
B. Image Recovery Algorithm
Detailed steps of reconstruction process at receiver end are presented in algorithm IMAGE_RECOVERY.
Algorithm: IMAGE_RECOVERY
1.
2.
3.

The values of parameters from shared images are collected.

From the equation by using those parameters and solve equations to get pixel values of different images.
Collect all the pixels of these images to get the original image.
V.

EXPERIMENTAL RESULTS

The proposed recovering algorithm requires shared image. If an intruder gathers k-1 shared images, it acquires
k-1 hyper planes to calculate the intersection point in a k - dimensional space. Result will be a line passing
through the point. This line contains many solutions but only one is correct. In experimental results we observe
that by selecting k =3 it contain over two thousand valid solution for each hyper plane constituting from 3 pixels.
In this experiment, images with size 256 256 pixels are used. It is observed that there exist over 20,000 hyper
planes and each hyper plane produces more than 2,000 valid solutions which leads to over 40,000,000 valid
reconstructed images for any brute force attack. This is almost impossible for any intruder to figure out original
protected image from such a large number of possibilities. Thus, experimental results and preliminary analysis
illustrate that the proposed approach is a simple but efficient method to share an image secretly.
The experimental results are shown in Fig. 1 and Fig. 2, where two reprehensive images are taken Fig 1(a) shows
the original input show the image shared images achieved by implementation of Blakley's scheme and Fig 1 (f)
is the reconstructed images using the recovery algorithm of Blakley.
Fig. 2(a)-(c) show the original three images to be transmitted using proposed scheme. The shared images using
this image are shown in Fig. 2(d) to Fig. 2(h) and output reconstructed images are show in Fig. 2(i) to Fig. 2(k).
It is evident from Fig. 2 that three images can simultaneously be transmitted securely and leaching to large
number possible images to be checked by any intruder. But the limitation of the proposed scheme is that if only

45

Ashutosh Gupta et al., World Applied Programming, Vol (2), No (1), January 2012.

one image is to be transmitted then algorithm includes two fictitious images of two additional copies of the same
image.
Thus, it is a better method to achieve secure transmissions of images. In this method nobody comes to know the
complete information about any of the image. If only one image to be transmitted, then two images can be taken.
Using proposed scheme, both flaw 1 and flaw 2 have been removed. We need only 3 images out of 5 shares for
reconstruction. In the proposed scheme, three images are sending at a time. If only one image is to be send by a
sender then two fake images can be used, for sending one image. Hence the time complexity increases but level
of security also increases. Therefore there is a trade-off between complexity and security.
VI.

CONCLUSION

This paper present the analysis of previous secret image sharing scheme and identify the flaws in that the
solution of the flaws in Blakley's secret sharing scheme are proposed. A new efficient sharing scheme is
proposed. The effectiveness of the proposed scheme has been presented through analysis and experimental
results. It is evident from the experimental results and analysis that proposed scheme is more efficient than the
existing scheme and can transmit and recover more than one image effectively.
.
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]

W. Zhao, R. Chellappa, A. Rosenfeld, and P. Phillips (2002) Face recognition: A literature survey. Technical Report CAR-TR-948,
UMD CS-TR-4167R, August, 2002
M. Turk and A. Pentland (1991) Eigenfaces for recognition J. Cognitive Neuroscience, 1991, 3(1), pp. 71-86.
A. Hyvrinen, J. Karhunen, E. Oja (2001) Independent Component Analysis Wiley,New York, 2001.
M. S. Bartlett and T. J. Sejnowski (1997) Independent components of face images: a representation for face recognition Proceedings of
the 4th Annual Jount Symposium on Neural Computation, Pasadena, CA, May 17, 1997.
Jian Yang, David Zhang and Jing-yu Yang (2005) Is ICA Signi_cantly Better than PCA for Face Recognition? Proceedings of the
Tenth IEEE International Conference on Computer Vision (ICCV'05) 1550-5499/05, 2005
D. Swets and J. Weng (1996) Using Discriminant Eigenfeatures for Image Retrieval IEEE Transactions on Pattern Analysis and
Machine Intelligence, vol. 18, pp. 831-836, 1996.
P. Belhumeur, J. Hespanha, and D. Kriegman (1997) Eigenfaces vs. Fisherfaces: Recognition Using Class Speci_c Linear Projection
IEEE Transaction on Pattern Analysis and Machine Intelligence, vol. 19, pp. 711-720, 1997.
K. Fukunaga (1989) Statistical Pattern Recognition New York: Academic Press, 1989
M. Gu, S.C. Eisenstat (1994) A Stable and Fast Algorithm for Updating the Singular Value Decomposition Research Report YALE
DCR/RR-996, 1994, Yale University ,New Haven, CT
S. Chandrasekaran, B.S. Manjunath, Y.F.Wang, J. Winkeler, and H. Zhang (1997) An Eigenspace update algorithm for image analysis
journal of Graphical Model and Image Processing, 1997
K. Etemad and R. Chellappa (1997) Discriminant Analysis for Recognition of Human Face Images journal of Optical Society of
America A, pp. 1724-1733, Aug.1997
W. W. Bledsoe (1966) The model method in facial recognition Panoramic Research Inc., Palo Alto, CA, and Rep. PRI: 15, Aug. 1966
T. Kanade (1973) Picture processing system by computer complex and recognition of human face Dept. of Information Science, Kyoto
University, Nov. 1973
A.L. Yuille, D.S. Cohen, and P.W. Hallinan (19889) Feature extraction from faces using deformable templates proc. CVPR, San Diego,
CA, June 1989.
M. Black and Y. Yacoob (1997) Recognizing facial expressions in image sequences using local parameterized models of image motion
Comput.Vis" vol. 25, no. 1, pp. 23-48, 1997.
M. Pantic and L. J. M. Rothkrantz (2000) Expert system for automatic analysis of facial expression Image Vis. Comput. J., vol. 18, no.
11, pp. 881-905, 2000.
C. Darwin (1965) The Expression of the Emotions in Man and Animals Chicago, IL: Univ. of Chicago Press, 1872, 1965.
M. Lewis and J. M. Haviland-Jones, Eds. Handbook of Emotions Guilford Press, New York, 2000, pp. 236-249.
P. Ekman and W. Friesen (1978) Facial Action Coding System Palo Alto, CA: Consulting Psychol. Press, 1978.
K. R. Scherer and P. Ekman, Eds. Handbook Methods in Non-Verbal Behavior Research Cambridge Univ. Press, Cambridge, MA,
1982.

46

Ashutosh Gupta et al., World Applied Programming, Vol (2), No (1), January 2012.

Figure 1. Input image to be transmitted (a) Original Image,(b) shared image 1, (c) shared image 2, (d) shared image 3, (e) shared image 4, (f) Shared Image 5
,(g)Reconstructed Image using Blakley's scheme.

Figure. 2. Input images used in proposed scheme (a) Original image 1, (b) Original image 2, (c) Original image 3.

47

Ashutosh Gupta et al., World Applied Programming, Vol (2), No (1), January 2012.

Figure. 3. Shared images used in proposed scheme (a) Shared image 1, (b) Shared image 2, (c) shared image 3, (d) Shared image 4, and (e) Shared image 5.

Figure. 4. Output images reconstructed using proposed scheme (a) Reconstructed Image1, (b) Reconstructed Image2, (c) Reconstructed Image3.

48