Loading [MathJax]/jax/output/SVG/jax.js
Research article

A global progressive image secret sharing scheme under multi-group joint management


  • Received: 17 September 2023 Revised: 10 December 2023 Accepted: 21 December 2023 Published: 26 December 2023
  • Diverging from traditional secret sharing schemes, group secret sharing schemes enable the recovery of secret information through collaborative efforts among groups. Existing schemes seldom consider the issue of the secrecy level of image information between different groups. Therefore, we propose a global progressive image secret sharing scheme under multi-group joint management. For inter-group relations, multiple groups with different priority levels are constructed using the approach of bit-polar decomposition. In this arrangement, higher-level groups obtain clearer secret image information. For intra-group relations, a participant-weighted secret sharing scheme is constructed based on Chinese Remainder Theorem and Birkhoff interpolation, in which the participants' secret sub-shares are reusable. During the recovery process, the sub-images can be recovered within the intragroup with the corresponding level. Groups collaborate through lightweight overlay operations to obtain different layers of secret images, achieving a global progressive effect. Analysis results show that the scheme is both secure and practical for group secret sharing.

    Citation: Lina Zhang, Jing Zhang, Jiaqi Sun, Qingpeng Chen. A global progressive image secret sharing scheme under multi-group joint management[J]. Mathematical Biosciences and Engineering, 2024, 21(1): 1286-1304. doi: 10.3934/mbe.2024055

    Related Papers:

    [1] Xingxing Jia, Yixuan Song, Daoshun Wang, Daxin Nie, Jinzhao Wu . A collaborative secret sharing scheme based on the Chinese Remainder Theorem. Mathematical Biosciences and Engineering, 2019, 16(3): 1280-1299. doi: 10.3934/mbe.2019062
    [2] Xuehu Yan, Longlong Li, Lintao Liu, Yuliang Lu, Xianhua Song . Ramp secret image sharing. Mathematical Biosciences and Engineering, 2019, 16(5): 4433-4455. doi: 10.3934/mbe.2019221
    [3] Xiaoping Li, Yanjun Liu, Hefeng Chen, Chin-Chen Chang . A novel secret sharing scheme using multiple share images. Mathematical Biosciences and Engineering, 2019, 16(6): 6350-6366. doi: 10.3934/mbe.2019317
    [4] Li-na Zhang, Jia-qi Sun, Xiao-yu Zhang, Qing-peng Chen, Jing Zhang . Two-level QR code scheme based on region matrix image secret sharing algorithm. Mathematical Biosciences and Engineering, 2023, 20(9): 16678-16704. doi: 10.3934/mbe.2023743
    [5] Feng Liu, Xuehu Yan, Lintao Liu, Yuliang Lu, Longdan Tan . Weighted visual secret sharing with multiple decryptions and lossless recovery. Mathematical Biosciences and Engineering, 2019, 16(5): 5750-5764. doi: 10.3934/mbe.2019287
    [6] Li-na Zhang, Chen-yu Cui, Xiao-yu Zhang, Wei Wu . Adaptive visual cryptography scheme design based on QR codes. Mathematical Biosciences and Engineering, 2022, 19(12): 12160-12179. doi: 10.3934/mbe.2022566
    [7] Yanjun Liu, Chin-Chen Chang, Peng-Cheng Huang . Security protection using two different image shadows with authentication. Mathematical Biosciences and Engineering, 2019, 16(4): 1914-1932. doi: 10.3934/mbe.2019093
    [8] Song Wan, Guozheng Yang, Lanlan Qi, Longlong Li , Xuehu Yan, Yuliang Lu . Multiple security anti-counterfeit applications to QR code payment based on visual secret sharing and QR code. Mathematical Biosciences and Engineering, 2019, 16(6): 6367-6385. doi: 10.3934/mbe.2019318
    [9] Ching-Chun Chang, Chang-Tsun Li . Algebraic secret sharing using privacy homomorphisms for IoT-based healthcare systems. Mathematical Biosciences and Engineering, 2019, 16(5): 3367-3381. doi: 10.3934/mbe.2019168
    [10] Xuehu Yan, Xuan Zhou, Yuliang Lu, Jingju Liu, Guozheng Yang . Image inpainting-based behavior image secret sharing. Mathematical Biosciences and Engineering, 2020, 17(4): 2950-2966. doi: 10.3934/mbe.2020166
  • Diverging from traditional secret sharing schemes, group secret sharing schemes enable the recovery of secret information through collaborative efforts among groups. Existing schemes seldom consider the issue of the secrecy level of image information between different groups. Therefore, we propose a global progressive image secret sharing scheme under multi-group joint management. For inter-group relations, multiple groups with different priority levels are constructed using the approach of bit-polar decomposition. In this arrangement, higher-level groups obtain clearer secret image information. For intra-group relations, a participant-weighted secret sharing scheme is constructed based on Chinese Remainder Theorem and Birkhoff interpolation, in which the participants' secret sub-shares are reusable. During the recovery process, the sub-images can be recovered within the intragroup with the corresponding level. Groups collaborate through lightweight overlay operations to obtain different layers of secret images, achieving a global progressive effect. Analysis results show that the scheme is both secure and practical for group secret sharing.



    With the rapid development of global network and communication technologies, we have entered the era of networking and informationization [1]. In this era, a vast number of images carrying personal information are rapidly spreading across the Internet. Even seemingly meaningless communication data can inadvertently leak important information. Therefore, protecting the confidentiality and integrity of image data has become an important research topics both domestically and internationally.

    Currently, image security techniques mostly include image information hiding, image encryption and image secret sharing. Among them, image secret sharing (ISS) [2], as a specific application direction of secret sharing scheme [3] in image information security protection, provides a higher level of security and confidentiality for image security protection. It applies the concept of secret sharing to the storage and transmission of image information, effectively solving the problem of image secret information leakage and tampering.

    ISS can be divided into two categories: Threshold secret sharing scheme (TSS) [4,5,6,7,8,9] and progressive image secret sharing scheme (PISS). However, some inherent defects have in existing TSS schemes. The size of the shares at least twice that of the secret image due to the high pixel expansion ratio. Also, the reconstructed image may exhibit low quality. To solve these problems, researchers have proposed many PISS schemes.

    The fundamental concept of PISS is to accomplish secret sharing by progressively unveiling various layers of an image. Liu and Yang [10] proposed three sharing models to construct a scalable image secret sharing scheme. Yan et al. [11] constructed a generalized progressive image secret sharing scheme with pixel non-expansion. Guo et al. [12] proposed a progressive image secret sharing scheme that divides the generated shared copies into multiple tiers, and there is a risk of partial reconstruction due to the fact that an attacker can try to recover the overall secret image information by using part of the acquired image information. To overcome this drawback, Bhattacharjee et al. [13] introduced a progressive image secret sharing scheme utilizing compression-aware generation of fixable shares. Expanding upon this groundwork, references [14,15] have proposed a progressive image secret sharing scheme based on compressive sensing (CS) and multi-level access control, respectively. In [14], the encoding and measurement of the image were implemented, while in [15], the secret image was divided into multiple layers with varying access permissions. Moreover, both [14] and [15] present techniques for gradually recovering different levels of the secret image, which provides valuable references for achieving secure image secret sharing.

    As the need for information sharing continues to grow, the traditional solution of sharing secret images by a single group does not meet the needs of multiple groups. In order to provide a more flexible sharing mechanism, group secret sharing schemes [16,17,18,19,20] have emerged. This type of scheme allows multiple groups to share secret images at the same time for broader information sharing. In [17], Lagrange interpolation and the Chinese Remainder Theorem were utilized to construct a secret sharing scheme, which is the first time that the concept of asymptotic secret sharing is extended to multi-group secret sharing schemes. Li et al. [18] employed the Chinese Remainder Theorem to construct a group secret sharing scheme, but it cannot share a large number of secrets efficiently. Yang et al. [19] constructed an image secret sharing scheme based on group cooperation using homomorphic Lagrangian interpolation, which realizes the sharing of secret images within and between groups, however, the number of secrets is limited to one. Wu et al. [20] proposed a group-based image secret sharing scheme that only allows for all-or-nothing recovery, and once a participant provides a dishonest share, the secret image will not be recovered.

    Existing group secret sharing schemes usually adopt an all-or-nothing approach to image secret sharing, which does not take into account the differences in the level and importance of image information, leading to inefficiencies in some cases. Secret image sharing is often accompanied by complex contexts, and group-based scheme [20] for all-or-nothing image recovery are inflexible and not applicable to complex application scenarios. Therefore, the diverse demands of various application scenarios necessitate the development of more adaptable progressive secret image sharing schemes, tailored to accommodate these evolving requirements.

    To ensure the security of different level information in the images, we propose a global progressive image secret sharing scheme based on multi-group joint management. Specifically, the present scheme utilizes bit-polar decomposition and priority assignment to give priority to groups, and the group with higher priority will receive a shared share of the image with higher pixel bits. Within each group, the Chinese Remainder Theorem is utilized to design an intra-group secret sharing scheme with participants having different weights. In the recovery phase, each group recovers the shared copies of the held images using Birkhoff interpolation and the Chinese Remainder Theorem, where Birkhoff interpolation can effectively fill in the missing image pixels. Intergroups utilize lightweight superposition operations for global progressive recovery of images. Even if a group is unable to fully recover an image, it is possible to recover the image gradually by cooperating with other groups, improving the efficiency of sharing.

    The sections of this paper are organized as follows. The preparatory knowledge is presented in Section 2, the overall scheme design is presented in Section 3, the analysis and proof of the scheme is in Section 4 and the conclusion of the paper is given in Section 5.

    The notations in Table 1 are used throughout this paper.

    Table 1.  Notations.
    Notation Description
    D Secret distributor
    S Secret image
    Gi Secret participating group
    ni The number of participants in Gi
    Gi,j The j-th participant in Gi
    Wi,j The weight of Gi,j
    ti The threshold of Gi
    Pi The sub-image after bipolar decoding.
    si,j The secret share of Gi,j
    Ii Image restoration of Gi

     | Show Table
    DownLoad: CSV

    Lakac et al. [10] originally proposed the concept of bit-level decomposition for image sharing. Let B(O)={bm1,,b2,b1,b0} be the set of bit-planes of a certain pixel in image O in Figure 1, where m is the depth of a pixel, and m belongs to {m|1m8,mZ+}. b0 is the least significant bit (LSB), and bm - 1 is the most significant bit (MSB). Based on the characteristics of a pixel value, the bits in B(O) follows a partial ordering bm1>bm2>...>b0, where bi>bj means that bi is more significant than bj, or simply i>j. In general, in a binary system, the changing of higher bits has a significant impact on the value, while the changing of lower bits has a smaller impact on the value. The partial ordering of the bits in B(O) by their significance implies the presence of priority among the bits. According to the desired group priority criteria, this solution assigns decomposed secret images at bit-level to each group and then shares image pixels with different bit depths.

    Figure 1.  The set of bit-planes of a certain pixel in image O.

    The priority decomposition divides the image O into n partitions P1,P2,,Pn satisfying Eq (1).

    {iB(Pi)=B(O)for  1inB(Pi)B(Pj)=ϕfor  1ijn (1)

    The secret sharing scheme based on Birkhoff interpolation uses a different polynomial interpolation method compared to Shamir's secret sharing scheme, which supports a higher number of polynomials, thus supporting a larger number of participants with higher efficiency and stronger security. Therefore, this scheme applies Birkhoff interpolation to the secret recovery process, which makes the scheme have better protection performance.

    Definition 1. Definition that X={x1,x2,...,xk}, where x1<x2<...<xk. E is a interpolation matrix with binary entries which is defined as E=(ei,j)1ik,0jl, where I(E)={(i,j):ei,j=1} and N=|I(E)|. C is a set of N real values which is defined as C={ci,j:(i,j)I(E)}.

    The corresponding Birkhoff interpolation problem [21] is <X,E,C> a problem of finding a polynomial P(X)RN1[x] satisfying N equations P(j)(xi)=ci,j(i,j)I(E). Where P(j)(xi) is the jth derivative of P(x) and RN1[x] is the set of all possible polynomials of order at most N1.

    Theorem 1. Let the Birkhoff interpolation problem corresponding to the triple <X,E,C> be will posed, where E satisfies Eq (2).

    t,(0tl):tj=0ki=1ei,j(t+1) (2)

    where l is the highest derivative order in the data and k is the number of interpolation points.

    Theorem 2. This interpolation problem (Definition 1) has a unique solution if the interpolation matrix E satisfies Theorem 1 and does not contain supported l-odd length sequences.

    Theorem 3. If the conditions in Theorem 2 and the following conditions co-exist, there exists a unique solution to the Birkhoff interpolation problem over the finite domain GF(q) as shown in Eq (3). where l is the highest order derivative in the data.

    q>2l+2×(l1)l12×(l1)!×x(l1)(l2)2k (3)

    In this paper, a global progressive image secret sharing scheme based on multi-group joint management is developed, as shown in Figure 2. The secret image S is classified into secret levels using the bit-polar decomposition technique, and then different groups G1, G2 andG3are assigned to the sub-images P1, P2 and P3 with different secret levels, respectively, such that each group is prioritized G1>G2>G3. As shown in Figure 2(a), G1 may utilize the group's sub-image P1 to generate an intra-group secret share s1,j to be assigned to each participant within G1. As shown in Figure 2(b), when the secret image is recovered, G1,j cooperate to fully recover P1, and participants within G2 and G3 can fully recover P2 and P3. As shown in Figure 2(c), when cooperating between groups, each Gi can make the clarity of the recovered image gradually increase by overlaying the sub-image Pi, which realizes the global progressive recovery effect.

    Figure 2.  Schematic diagram of the scheme.

    Given a secret image S of size m×r. The scheme contains a trusted secret distributor D, a bulletin board and three participant groups G1, G2 and G3. D can post and update content on the bulletin board, and others can only read or download it. G1 consists of n1 participants, and so on. Gi consists of ni participants, n=n1+n2+n3. Gi,j denotes the j-th participant of the i-th group, and the weight of each participant is Wi,j. In addition, each group has a corresponding threshold requirement, such that the threshold for G1 is t1, and the threshold for Gi is ti. The corresponding sub-image can be reconstructed only if the sum of the weights of the participants in the group satisfies the threshold of the corresponding group. The sub-secrets of the participants in the group in this scheme are selected and saved by themselves and can be used multiple times.

    D chooses two large prime numbers p and q at random and computes the product N=pq, satisfying the attacker that it is computationally infeasible to derive the sum p is q knowing N. D picks a random g in [N,N] satisfying g(p1)(q1)2=1modN. Pick s0 at random in [2,N] such that s0 and (p1)(q1) are mutually prime, compute p0=gs0modN, and compute the smallest positive integer h such that s0h=1modφ(N), which will keep s0 secret and { g,N,p0,h} public.

    For convenience, this section divides the secret image S into 3 different quality sub-images P1, P2 and P3, as shown in Eq (4). The higher the hierarchy of the sub-image, the more information it contains. Distribute sub-images Pi to Gi, where priority G1>G2>G3. In the image reconstruction pahse, the secret image can be reconstruction progressively by using sub-images from different groups.

    {B(P1)={b7,b6}B(P2)={b5,b4}B(P3)={b3,b2,b1,b0} (4)

    During the image preprocessing, the secret image generates authorized sub-images for different groups. During the pixel allocation process, D uses the pixel values of the sub-image as coefficients of a polynomial. Participants randomly select sub-secrets and perform computations using these sub-secrets to generate public identifiers. D verifies the participant identities to ensure that each participant's identity is unique. Once the uniqueness of the participant's identity is verified, D distributes to each participant a share of the computed secrets, which are generated based on a polynomial computation, to ensure the security of the secrets. The algorithm process is shown in Algorithm 1.

    Algorithm1. Secret sharing algorithm
    Input: Sub-image Pi.
    Output: Secret share Si,j=yi,j(k).
    Step1: Iterate over each row of pixel ai in the sub-image Pi in turn. Constructing polynomials f(x)=a0+a1x+a2x2++ar1xr1.
    Step2: Gi,j randomly selects xi,j as the private key to be saved, calculates the public identity yi,j=gxi,jmodN, and sends yi,j to D.
    Step3: After receiving yi,j, D assesses the public identity identifiers held by any two different participants Gi,j and Gi,k. If yi,j=yi,k, an alarm is triggered, and both parties are required to resend. If yi,jyi,k, proceed to step 4.
    Step4: D computes Hi,j(x)=f(x)mod(xyi,j)wi,j.
    Step4: D computes Hi,j(x)=h0+h1x+h2x2++hni1xni1 and distributes Si,j=Hi,j(x) as a secret share to the Gi,j.

    By collaborating between participating groups, the secret share generation process constructed using Algorithm 1 is shown in Figure 3.

    Figure 3.  Secret share generation diagram.

    Secret recovery consists of two parts: Intra-group recovery and inter-group recovery. Gi,j hold their own secret shares si,j and restore the corresponding sub-image Pi by using the Chinese Remainder Theorem and Birkhoff interpolation, revealing partial information. During inter-group recovery, any group overlays their own sub-images Pi, and the image corresponding to pixel depth will be restored, while the remaining bit depths will remain in a noisy state. The algorithm process is shown in Algorithm 2.

    Algorithm2. Secret reconstruction algorithm
    Input: Si,j=Hi,j(x).
    Output: Reconstruction Image Ii.
    Step1: Identity verification process. Participant Gi,j calculates Ai,j=p0xi,j. D verifies if Ai,jh=yi,j based on the public identity identifiers yi,j and h of the participants. If Ai,jh=yi,j, then Gi,j is considered not deceiving, and the identity verification is passed. If Ai,jhyi,j, an alarm will be sent to Gi,j, requesting a resend.
    Step2: From the Si,j of Gi,j, the coefficients of Hi,j(x) can be found according to the Birkhoff interpolation, which leads to Hi,j(x).
    Step3: D calculates tij=1wijti and determines whether it is valid. If tij=1wijti, this means that the set of participants intending to participate in the recovery are qualified to participate in restoring the sub-image of their respective group. If tij=1wij<ti, it indicates that they are not qualified to participate in the recovery.
    Step4: Under the condition that the identity of the participant is confirmed to be true and tij=1wijti is judged to be valid, the Chinese Remainder Theorem is applied to uniquely solve Y=ni=1riMibi(modM), where M=ni=1mi, bi=M1i(modmi) and the only polynomial f(x)=a0+a1x+a2x2++ar1xr1 with a number of times of r1 is computed according to H(x)=f(x)mod(xyi,j)wi,j, and the group can be recovered by the pixel value of the corresponding sub-image Pi.
    Step5: By overlaying the recovered sub-images from each group in the inter-group operation, the reconstruction image Ii has a globally progressive recovery effect.

    Using Algorithm 2 for intra-group and inter-group image restoration, the restoration process is shown in Figure 4.

    Figure 4.  Image reconstruction schematic.

    The identity verification step described in Section 3.4 of this paper, constant Eq (5) holds.

    Ai,jh=yi,j (5)

    Proof. During the secret recovery phase, it is not possible for Gi,j to provide fake shares Ai,j to D. Because D verifies whether the equation Ai,jh=yi,j holds after receiving the shares Ai,j provided by Gi,j, where Ai,jh=p0xi,jh=gs0xi,jh=(gs0h)xi,j=gxi,j=yi,j. Only proceeds to the next step if the verification passes.

    In this paper, in the secret recovery algorithm in Section 3.4, the secret share Si,j=Hi,j(x), the secret image can be recovered efficiently using the designed Algorithm 2.

    Proof. The secret share Si,j=Hi,j(x) held by the authenticated Gi,j is derived from the coefficients of Hi,j(x) based on Birkhoff interpolation, which leads to Hi,j(x). Then applying the Chinese Remainder Theorem, calculate f(x) (The coefficients of f(x) are the pixel values of the corresponding region of the recovered image).

    1) Find Hi,j(x) based on the secret share Si,j=Hi,j(x).

    The basic idea of Birkhoff interpolation is to construct an interpolation polynomial based on known data points, which can approximate and interpolate between these points. The key of this method is to transform the known data points into a coefficient matrix and solve for the coefficients of the polynomial through matrix operations. As shown in Eq (6), this scheme constructs a vector S=[S(x0),S(x1),...,S(xr)] consisting of the secret shares Si,j of Gi,j. Construct a Vandermonde matrix V by arranging the pairs (xi,S(xi)) as rows or columns of the matrix. Where r is the number of pixel points, ni is the degree of the polynomial Hi,j(x) being solved, and Gi is the number of participants involved. Assuming that the coefficient matrix of the polynomial Hi,j(x) is C, solves for C in C=V1S, and then obtain Hi,j(x) by solving the equation.

    V=[1x1x12...x1ni1x2x22...x2ni...............1xrxr2...xrni] (6)

    2) After obtaining Hi,j(x), f(x) can be solved by the Chinese Remainder Theorem through H(x)=f(x)mod(xyi,j)wi,j, where the coefficients of f(x) correspond to the pixel values of the image in the corresponding region.

    The Chinese Remainder Theorem is a mathematical method used to solve a system of congruent equations. According to the congruent equation H(x)=f(x)mod(xyi,j)wi,j, obtain f(x)=H(x)mod(xyi,j)wi,j. Therefore, by solving Hi,j(x) modulo (xyi,j)wi,j, determine the function f(x). First, calculate the modulus m. According to the given condition, m=(xyi,j)wi,j. Then, calculate the value of the function Hi,j(x) and perform modulo m operation on it, which is H(x)modm.

    According to the mathematical definition, H(x)modm=H(x)m×floor(H(x)m), where floor(H(x)m) represents the floor of H(x)m. By substituting m=(xyi,j)wi,j into the equation, get H(x)modm=H(x)(xyi,j)wi,j×floor(H(x)(xyi,j)wi,j)=f(x), where the coefficient of f(x) represents the pixel value of the corresponding region in the restored image.

    1) Protects against external attacks. It is infeasible for an attacker to try to derive a sub-secret xi,j of Gi,j through the open yi,j.

    Proof. It is not possible for an attacker to download the information yi,j of Gi,j from the bulletin board and try to obtain the participant's sub-secret xi,j. Due to the difficulty of solving discrete logarithms, it is computationally infeasible to calculate xi,j given yi,j and g, as yi,j=gxi,jmodN.

    2) Reusable sub-secrets. When it is necessary to reassign the secret image, D randomly selects s0 from [2,N] again and constructs a new polynomial f(x)=a0+a1x+a2x2++ar1xr1 with degree r-1. Meanwhile, D calculates H(x)=f(x)mod(xyi,j)wi,j for each participant and updates their secret shares si,j without changing their sub-secrets xi,j, so each participant can use their sub-secret multiple times.

    3) During the recovery process, when the number of participants in Gi does not satisfy the threshold condition ti, Gi may not recover the corresponding sub-image to ensure the security of the scheme.

    Proof. In this scheme, to recover the shared sub-image Pi, it is necessary to reconstruct an r-1 degree polynomial f(x)=a0+a1x+a2x2++ar1xr1. During the secret recovery process, it is required that the sum of weights of participants in Gi,j reaches the threshold value ti, in order to guarantee accurate calculation of m=(xyi,j)wi,j. Then, the r-1 degree polynomial f(x)=a0+a1x+a2x2++ar1xr1 can be further calculated using the pixel values of the sub-image Pi. For sub-images Pi with different priorities, if the sum of weights of the participants in Gi,j does not reach the threshold value ti, it will not be possible to calculate m=(xyi,j)wi,j, and hence the correct r-1 degree polynomial f(x)=a0+a1x+a2x2++ar1xr1 cannot be obtained, resulting in unsuccessful image recovery in the group recovery process of the image Ii. In this case, the cooperation between this authorized group and other groups cannot successfully recover the image Ii+j, and hence it is not possible to perform inter-group recovery.

    This chapter presents the experimental results of our proposed approach, and we evaluate the effectiveness of the scheme through experiments and analysis.

    This section provides an example of this scheme to illustrate the generation of secret shares and the reconstruction of images. This section verifies the proposed scheme by taking, for example, three groups, and suppose there are three separate groups Gi(i=1,2,3), where the priority of G1>G2>G3. The secret message obtained from G1 is the clearest, while the secret message obtained from G3 is the closest to black, making the secret message almost unobservable through the visual system. Figure 5(a) serves as the secret image, and Figure 5(b)-(d) shows each group of sub-image. Figures 6-8 show the secret shares sent to participants within the G1,G2,G3.

    Figure 5.  (a) Secret image, (b)-(d) Sub-images distributed to G1-G3.
    Figure 6.  Secret shares of G1,j.
    Figure 7.  Secret shares of G2,j.
    Figure 8.  Secret shares of G3,j.

    The security of secret image in this scheme is demonstrated by experimental results. In Figure 9(a), when the secret shares of the 3 participants, denoted as G1,j, do not satisfy the threshold condition, the reconstructed image appears as a random pattern. However, in Figure 9(b), when the secret shares of the 4 participants, also denoted as G1,j, meet the threshold condition, the corresponding sub-images can be successfully reconstructed. In other words, only the participants who satisfy the threshold condition can participate in the effective image recovery process, while those who do not satisfy the threshold condition cannot be involved in image reconstruction.

    Figure 9.  (a) Three participants from G1 participated in the recovery; and (b) Four people from G1 participated in the recovery.

    After obtaining the corresponding sub-images Pi, cooperation between different groups was carried out to recover the secret image. Specifically, in Figure 10(a)-(c), the image restoration process was completed through cooperative efforts between the pairs G2+G3, G1+G3 and G1+G2, respectively. It is worth noting that these cooperative methods are determined based on the selection of participants. In addition, in Figure 10(d), three groups, G1+G2+G3, participated in the restoration process simultaneously, and the secret image was successfully reconstructed. This result indicates that cooperation among multiple groups can significantly improve the efficiency and robustness of image restoration.

    Figure 10.  (a)-(c) Two groups participated in the recovery; and (d) Three groups participated in the recovery.

    The groups involved in the recovery are G1,G2 and G3, where the priority is G1>G2>G3. During the recovery process, the group with higher priority will restore a higher quality image. As shown in Figure 10(b), (c), the restored images I1+2 and I1+3, which were restored through the collaboration of G2 and G3, respectively with G1, demonstrate clear outlines but fail to fully recover and have lost some details. Additionally, G2 has higher priority over G3. Under the premise of G1 participation, the image I1+2, restored through the collaboration of G1 and G2, is clearer than the image I1+3, restored through the collaboration of G1 and G3. Due to the lower priority of G2 and G3, their collaborative restoration results in a less prominent feature in the restored image I2+3, as illustrated in Figure 10(a). When all three groups, G1, G2 and G3, participate in the restoration simultaneously, as shown in Figure 10(d), the restored image I1+2+3 exhibits the best image quality. At this point, the detailed information in the image has been nearly fully restored, resulting in a higher level of clarity and accuracy in the overall image.

    During the secret restoration process, different collaborative groups have varying priorities and restoration effects. The quality of the restored image is influenced by the number of participating groups and their priorities. As the number of participating groups gradually increases, the clarity and accuracy of the secret image improves, gradually moving away from its initial blurry state. When the number of groups is the same, the degree of improvement in clarity depends on the priority assigned to each group, with higher-priority groups helping to obtain clearer image information. In particular, when collaborating groups have a higher collective priority, the clarity and accuracy of the restored secret image are enhanced. Optimizing collaborative restoration strategies can significantly improve the quality and accuracy of the restored image.

    Image data is inevitably affected by various subjective and objective factors that cause pixel interference during transmission. To verify the impact of pixel interference on secret images before recovery, this section uses "Lena" and "Peppers" with a size of 264 × 173 as secret images, as shown in Figure 11. The results of the recovery process are shown in Figures 12 and 13. The recovered images were also tested after adding salt-and-pepper noise with a strength of 2% to the secret shares, as shown in Figures 14 and 15.

    Figure 11.  The secret image.
    Figure 12.  (a)-(c) Two groups participated in the recovery; and (d) Three groups participated in the recovery.
    Figure 13.  (a)-(c) Two groups participated in the recovery; and (d) Three groups participated in the recovery.
    Figure 14.  Recovered image after intensity 2% noise attack.
    Figure 15.  Recovered image after intensity 2% noise attack.

    Since the naked eye system has limitations in determining the quality of image restoration, precise computational methods are needed to measure the quality of image restoration. This section uses the mean square error (MSE) and peak signal to noise ratio (PSNR) to analyze the image recovery performance under noise.

    By analyzing experimental data, when the MSE value is smaller and the PSNR value is larger, the reconstructed image is more similar to the original image and the difference between the two is smaller. When the PSNR is higher than 40 dB, it indicates that the quality of the recovered image is extremely good and very close to the secret image. When the PSNR is between 30~40 dB, it indicates that the image quality is good. The MSE and PSNR between two given images P1 and S, with sizes of m×n, are defined as Eqs (7) and (8).

    MSE=1m×nmi=1nj=1|P1(i,j)S(i,j)|2 (7)
    PSNR=10log10(2552MSE) (8)

    According to Tables 2 and 3, it can be observed that with inter-group collaboration, the PSNR of the recovered images is between 30 dB and 59 dB, while the MSE remains between 0.007 and 0.010. This indicates that the higher the priority of Gi involved in the recovery, the better the quality and PSNR of the corresponding recovered image. With the increase in noise attack intensity and cooperation group, it is inevitable that there will be some impact on the reconstruction of the secret, but it is within the visible range. Therefore, this scheme exhibits good robustness to noise attacks.

    Table 2.  MSE comparison of the secret images before and after recovery.
    Image MSE
    I1 I2 I3 I1+2 I1+3 I2+3 I1+2+3
    School 0.0086762 0.0101109 0.0102801 0.0083329 0.0085976 0.0092178 0.0077101
    Lena 0.0092364 0.0097547 0.0098518 0.0084010 0.0085517 0.0094234 0.0081386
    Peppers 0.0088518 0.0089732 0.0096518 0.0082565 0.0086374 0.0088647 0.0078360

     | Show Table
    DownLoad: CSV
    Table 3.  PSNR comparison of the secret images before and after recovery.
    Image PSNR
    I1 I2 I3 I1+2 I1+3 I2+3 I1+2+3
    School 47.7 31.6 27.5 58.0 57.6 46.3 58.3
    Lena 49.4 35.6 22.0 57.5 54.0 48.1 58.1
    Peppers 47.6 35.9 29.3 58.8 56.6 44.0 59.0

     | Show Table
    DownLoad: CSV

    The comparison between some related progressive image secret sharing schemes and the proposed scheme is shown in Table 4.

    Table 4.  Comparison of this scheme with the existing ISS schemes.
    Scheme Security level Identity verification Sub-secret reusable Participant characteristics Inter-group characteristics Intra-group characteristics Progressive
    [9] High No No One group No Equally No
    [12] Low No No One group No The importance of difference Yes
    [13] High No No One group No Equally important Yes
    [14] High No No One group No Equally Yes
    [15] Low No No One group No Equally Yes
    [20] High No No Multiple group No prioritization Equally important No
    This scheme High Yes Yes Multiple group Priority
    exists
    The importance of difference Yes

     | Show Table
    DownLoad: CSV

    First, there is a comparison of the level of security, as shown in the first column of Table 4. This scheme is superior to the scheme of Guo et al. [12] in terms of safety. It embeds secret pixel values into all coefficients of a polynomial, which may lead to unauthorized participants recovering the secret image, thus resulting in lower security. In the proposed scheme in this article, the secret image is divided into sub-images with different priorities, and secret sharing operations are performed on these sub-images, thus enhancing the security level of the secret image.

    Second, there is a comparison of the features used to verify the identity information of participants, as shown in the second column of Table 4. References [9,12,13,14,15,20] does not verify the identity of participants, which cannot prevent internal fraud, allowing adversaries to infiltrate and obtain information about the secret image using incorrect information. In this scheme, D verifies the identity information of participants by using the publicly available yi,j to calculate Ai,jh=yi,j.

    Third, as shown in the third column of Table 4, in this scheme, the secret share of each participant can be utilized for multiple secret sharing processes without necessitating any updates. References [9,12,13,14,15,20] has no reusable sub-secrets, and the secret shares need to be redistributed during each secret sharing process.

    Fourth, there is a comparison of the characteristics of participants, as shown in the fourth column of Table 4. The participants in references [9,12,13,14,15] belong to a group, so the secret image is managed by a single group. In real life, secret images need to be managed by multiple groups. Therefore, the scheme in this paper is more suitable for visual passwords.

    Fifth, only in this scheme there is prioritization among groups. In fact, it is unlikely that multiple groups co-managing mutual trade-offs can achieve complete equality in the true sense, and the priority assignment in this paper plays a decisive role in achieving global progressive recovery.

    Sixth, the shares generated by [13,20] have equal importance. Typically, individuals within a group are further classified into management members and regular members. Different access levels to classified information are assigned to members based on their importance levels. As a result, the proposed scheme in this paper is more versatile. In this scheme, each group has a specific threshold. When the sum of participant weights reaches the threshold, the corresponding sub-image can be restored. Compared to traditional (k,n) threshold schemes, this solution offers enhanced security.

    Finally, as shown in Table 4, Xie et al. [9] proposed a low-cost CS based multi-party secret image sharing scheme that enhances security through permutation and blurring, but it is not progressively recoverable. Guo et al. [12] resolved the issue of achieving only fully recovered or unrecovered images, and proposed non-expandable hierarchical shadowing for reconstructed images. Both [14] and [15] implemented a hierarchical structure, but without scalability. Wu et al. [20] is a joint multi-group image secret sharing scheme, but the reconstructed images are either fully recovered or not recovered at all. This scheme combines their strengths to achieve both hierarchy and progression, and secret images can be divided into arbitrary levels according to actual needs.

    In this paper, we introduce a novel global progressive image secret sharing scheme that combines group secret sharing with a progressive secret sharing system. The proposed scheme utilizes multi-group joint management to allow for global progressive recovery with varying priorities. Throughout the sharing process, the scheme employs bit-polar decomposition to divide the depth of the secret image's pixels, enabling groups to establish diverse priorities. During the recovery process, Birkhoff interpolation ensures the recovery of secret images within groups. Using lightweight superposition operations, distinct layers of secret images can be recovered with global progressivity among groups. In addition, the reuse of participant secret shares can effectively prevent external attacks. The scheme, as presented in this paper, finds potential application in transmitting product design diagrams. Future research will focus on detecting potential spoofing on the distributor's end and exploring methods for decentralizing the distributor's rights based on the current scheme.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    The authors thank the anonymous reviewers for their valuable comments. This work was supported by Xi'an Science and Technology Plan Project (No. 22GXFW0063); Shaanxi Provincial Department of Science and Technology Youth Project (Nos. 2021JQ-575 and 2021JQ-576); Shaanxi Provincial Department of Education Project (No. 19JK0526).

    The authors declare there is no conflict of interest.



    [1] Z. Yu, H. Gao, X. Cong, N. Wu, H. H. Song, A survey on cyber-physical systems security, IEEE Internet Things J., 10 (2023), 21670–21686. https://doi.org/10.1109/JIOT.2023.3289625 doi: 10.1109/JIOT.2023.3289625
    [2] P. Sarosh, S. A. Parah, G. M. Bhat, Utilization of secret sharing technology for secure communication: a state-of-the-art review, Multimedia Tools Appl., 80 (2021), 517–541. https://doi.org/10.1007/s11042-020-09723-7 doi: 10.1007/s11042-020-09723-7
    [3] A. Shamir, How to share a secret, Commun. ACM, 22 (1979), 612–613. https://doi.org/10.1145/359168.359176 doi: 10.1145/359168.359176
    [4] M. Naor, A. Shamir, Visual cryptography, in Advances in Cryptology — EUROCRYPT'94, (1995), 1–12. https://doi.org/10.1007/BFb0053419
    [5] W. Liu, Y. Xu, J. Chen, C. Yang, A (t, n) threshold quantum visual secret sharing, Int. J. Sens. Netw., 33 (2020), 74–84. https://doi.org/10.1504/IJSNET.2020.107863 doi: 10.1504/IJSNET.2020.107863
    [6] C. Thien, J. Lin, Secret image sharing, Comput. Graphics, 26 (2002), 765–770. https://doi.org/10.1016/S0097-8493(02)00131-0 doi: 10.1016/S0097-8493(02)00131-0
    [7] L. Tan, Y. Lu, X. Yan, L. Liu, L. Li, Weighted secret image sharing for a (k, n) threshold based on the Chinese Remainder Theorem, IEEE Access, 7 (2019), 59278–59286. https://doi.org/10.1109/ACCESS.2019.2914515 doi: 10.1109/ACCESS.2019.2914515
    [8] D. M. Huang, H. F. Xu, Q. He, Y. L. Du, Q. M. Wei, A secret sharing-based authentication scheme for ocean remote sensing images in cloud environment, Comput. Eng. Sci., 39 (2017), 1410–1418. https://doi.org/10.3969/j.issn.1007-130X.2017.08.004 doi: 10.3969/j.issn.1007-130X.2017.08.004
    [9] D. Xie, B. Wu, F. L. Chen, T. C. Wang, Z. B. Hu, Y. B. Zhang, A low-overhead compress sensing‑driven multi‑party secret image sharing scheme, Multimedia Syst. , 29 (2023), 1187–1202. https://doi.org/10.1007/s00530-023-01049-2 doi: 10.1007/s00530-023-01049-2
    [10] Y. Liu, C. Yang, Scalable secret image sharing scheme with essential shadows, Signal Process. Image Commun., 58 (2017), 49–55. https://doi.org/10.1016/j.image.2017.06.011 doi: 10.1016/j.image.2017.06.011
    [11] X. Yan, Y. Lu, L. Liu, A general progressive secret image sharing construction method, Signal Process. Image Commun., 71 (2019), 66–75. https://doi.org/10.1016/j.image.2018.11.002 doi: 10.1016/j.image.2018.11.002
    [12] C. Guo, C. C. Chang, C. Qin, A hierarchical threshold secret image sharing, Pattern Recognit. Lett., 33 (2012), 83–91. https://doi.org/10.1016/j.patrec.2011.09.030 doi: 10.1016/j.patrec.2011.09.030
    [13] T. Bhattacharjee, S. P. Maity, S. R. Islam, Hierarchical secret image sharing scheme in compressed sensing, Signal Process. Image Commun., 61 (2018), 21–32. https://doi.org/10.1016/j.image.2017.10.012 doi: 10.1016/j.image.2017.10.012
    [14] D. Xie, H. J. Zhu, F. L. Chen, B. Wu, J. H. Yang, A compressed sensing-based progressive secret image sharing scheme and security analysis, Digital Signal Process., 143 (2023), 104273. https://doi.org/10.1016/j.dsp.2023.104273 doi: 10.1016/j.dsp.2023.104273
    [15] X. L. Wang, D. Xie, F. L. Chen, B. Wu, Y. Y. Zen, Progressive and multi-level secret image sharing scheme with hierarchical shadows, Multimedia Tools Appl., 81 (2022), 31039–31059. https://doi.org/10.1007/s11042-022-12951-8 doi: 10.1007/s11042-022-12951-8
    [16] R. Xu, X. Wang, K. Morozov, Group authentication for cloud-to-things computing: review and improvement, Comput. Networks, 198 (2021), 108374. https://doi.org/10.1016/j.comnet.2021.108374 doi: 10.1016/j.comnet.2021.108374
    [17] K. Meng, F. Miao, W. Huang, Y. Xiong, Tightly coupled multi-group threshold secret sharing based on Chinese Remainder Theorem, Discrete Appl. Math., 268 (2019), 152–163. https://doi.org/10.1016/j.dam.2019.05.011 doi: 10.1016/j.dam.2019.05.011
    [18] H. X. Li, L. J. Pang, W. D. Cai, An efficient threshold multi-group-secret sharing scheme, in Fuzzy Information and Engineering, (2007), 911–918. https://doi.org/10.1007/978-3-540-71441-5_99
    [19] C. N. Yang, X. Wu, H. Y. Lin, C. Kim, Intragroup and intergroup secret image sharing based on homomorphic Lagrange interpolation, J. Inf. Secur. Appl., 61 (2021), 102910. https://doi.org/10.1016/j.jisa.2021.102910 doi: 10.1016/j.jisa.2021.102910
    [20] Z. Wu, Y. Liu, X. Jia, A novel hierarchical secret image sharing scheme with multi-group joint management, Mathematics, 8 (2020), 448. https://doi.org/10.3390/math8030448 doi: 10.3390/math8030448
    [21] W. G. Ge, Secret sharing scheme based on Birkhoff interpolating polynomial, J. China Acad. Electron. Inf. Technol., 13 (2018), 170–173. https://doi.org/10.3969/j.issn.1673-5692.2018.02.011 doi: 10.3969/j.issn.1673-5692.2018.02.011
  • This article has been cited by:

    1. Yu-Ru Lin, Justie Su-Tzu Juan, RG-Based Region Incrementing Visual Cryptography with Abilities of OR and XOR Decryption, 2024, 16, 2073-8994, 153, 10.3390/sym16020153
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1439) PDF downloads(52) Cited by(1)

Figures and Tables

Figures(15)  /  Tables(4)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog