Digital images are among the most communicated multimedia data types. Many of these images include private data that require a high level of security. The traditional image security schemes rely on cryptographic solutions to ensure the confidentiality or the authentication of image contents, and to ensure that the encryption key is not compromised. However, the continuous evolution of the attackers’ capabilities is making it harder than ever to achieve the goal of safeguarding the private data against breaches. Moreover, the centralization aspect of images’ storage makes them prone to availability attacks. In this paper, we propose a distributed and secure storage scheme for images, based on the Modified Information Dispersal Algorithm (MIDA), and taking into consideration the trade-off between the high security level and the associated computational overhead. The proposed solution applies block permutation on the image to ensure data confidentiality and then, divides it into k fragments that are encoded using the proposed parallel modified IDA. The output consists of n encoded fragments, instead of k, to ensure data availability. Next, each encoded fragment is authenticated using a lightweight Message Authentication Algorithm (MAA) to ensure data integrity with source authentication. Finally, the encoded fragments are distributed over n storage nodes (or multi-cloud providers). The resilience degree of such redundancy is (n − k), since only k fragments are required to reconstruct the original images. All the cryptographic steps such as permutation, IDA encoding and MAA consist of simple operations and they are based on a dynamic key. This ensures a high level of security since in each session, a new key is used to produce different cryptographic primitives as well as the update primitives, which are used to update the permutation and selection tables. The implementation results show that the proposed scheme meets the desired cryptographic properties to guard against different attacks. Finally, the performance tests show that the proposed scheme is lightweight with low overhead in terms of computations, communication and storage.

数字图像是通信最多的多媒体数据类型之一。这些映像中有许多包含要求高度安全性的私有数据。传统的图像安全方案依赖于加密解决方案，以确保图像内容的机密性或身份验证，并确保加密密钥不受影响。但是，攻击者功能的不断发展使实现保护私有数据免遭破坏的目标比以往更加困难。此外，图像存储的集中化方面使它们易于受到可用性攻击。在本文中，我们提出了一种基于改进信息分散算法（MIDA）的分布式安全图像存储方案，并考虑到高安全级别和相关的计算开销之间的权衡。所提出的解决方案对图像应用块置换以确保数据机密性，然后将其分为使用建议的并行修改的IDA编码的k个片段。输出由n个编码片段而不是k组成，以确保数据可用性。接下来，使用轻量级消息身份验证算法（MAA）对每个编码的片段进行身份验证，以确保通过源身份验证实现数据完整性。最后，编码的片段分布在n个存储节点（或多云提供商）上。这种冗余的弹性度为（n - k），因为只有k需要碎片来重建原始图像。所有加密步骤（例如置换，IDA编码和MAA）都由简单的操作组成，并且它们都是基于动态密钥的。这确保了高级别的安全性，因为在每个会话中，新密钥都用于生成不同的加密原语以及更新原语，而更新原语用于更新排列表和选择表。实施结果表明，所提出的方案符合期望的密码学性质，可以防范不同的攻击。最后，性能测试表明，所提出的方案是轻量级的，在计算，通信和存储方面具有较低的开销。