Secure and Efficient Image Retrieval over Encrypted Cloud Data

Abstract

This paper proposes a novel image retrieval scheme over encrypted cloud data, which achieves high efficiency and confidentiality. For the purpose of improving search efficiency, an index tree is often deployed in the image retrieval scheme. Meanwhile, the confidentiality of the sensitive cloud data, such as outsourced images, index tree, and query request, is also a key issue. Firstly, a balanced binary clustering algorithm is exploited over the integrated image features composed of basic features, such as HSV histogram and DCT histogram, yielding a balanced binary tree (BBT). In particular, due to the adoption of a balanced index tree, our scheme can achieve logarithmic search time. Secondly, the secure inner product is employed to encrypt the index vector and query feature. Finally, to resist the statistical attack of the frequency distribution of the retrieved results, we copy the database and merge the subtree of encrypted BBT to blind the search results. Security analysis and experimental results show that the proposed scheme is secure and efficient.

Privacy preserving search index for image databases based on SURF and order preserving encryption

“…They yield variably accurate results with a trade‐off between accuracy and computational complexity. Some of these studies use aggregation of feature vectors as in [16] to improve the accuracy by gathering more information about an image. The encryption mechanism in [17, 18] is based on hashing functions, while secure ASPE is employed in [16], and OPE is used in our work.…”

“…Authors of [17, 18] use the LSH technique in BOVW creation. On the other hand, in [16], a balanced binary clustering technique (BBC) is used to construct the index tree. We used k‐means++ clustering algorithm for its effective initial seeds selection mechanism and recommended performance.…”

“…Similarity measures highly affect the performance of image retrieval systems. In the aforementioned papers, the Euclidean distance is used in [18], EMD is used in [17] and the inner product similarity measure is employed in [16]. On the other hand, we used Jaccard's similarity.…”

“…It is noteworthy that the platforms used in the papers included in our comparative study are quite similar to ours. As for the scheme in [16], it employs a variant of the traditional BOVW model, where global features describing the whole image are used. These feature vectors are thus longer than those used to describe a traditional visual word representing a local feature of an image.…”

“…proposed a CBIR system based on a balanced binary tree of visual words (vocabulary tree) to accelerate the search. They used ASPE for encryption [16]. In their approach, they used a linear combination of the HSV histogram and the discrete cosine transform (DCT) histogram as image descriptors in constructing the vocabulary tree.…”