A privacy preserving image content retrieval method

which combines privacy preserving SIFT extraction method in encrypted image with binary SIFT algorithm is proposed. In this way

it can not only guarantee the image uploaded to the server is encrypted

but also keep its semantic privacy in encrypted domain. Image is encrypted by using Paillier homomorphic encryption

this can ensure security of image on server end. And SIFT features are extracted in the encrypted domain

after that the features are converted to a binary representation

reduce storage space and computational complexity. Experimental results show that the binary SIFT extracted from encrypted domain performs well in robustness evaluation

and the similarity distances between binary SIFT generated from the original image and the binary SIFT generated from the encrypted image keep the same

and ensure accuracy of image searching and matching in plaintext and ciphertext domain

and the matching efficiency is improved.