Secure and Traceable Multikey Image Retrieval in Cloud-Assisted Internet of Things

Tengfei Yang, Yuanyuan Li, Jiawei He, Zhiquan Liu, Fang Ren, Teng Wang, Gaopan Hou

Abstract: The privacy-preserving image retrieval technology permits users to retrieve outsourced images in a secure manner in cloud-assisted Internet of Things (IoT) environment. However, most of the existing schemes still have some blemishes, such as low performance, shared key, and untraceable malicious users. To this end, we present a secure and traceable multikey image retrieval, named as secure and traceable multikey image retrieval (STMIR). First, we design a novel privacy-preserving Mahalanobis distance comparison method (PPMDC) based on the learning with errors technology and Mahalanobis distance. And STMIR extracts image features utilizing the convolutional neural network (CNN) model to improve retrieval accuracy. Then, STMIR employs extracted image features, PPMDC and key conversion technology to achieve secure image retrieval that supports the multikey setting. Meanwhile, STMIR uses encrypted image watermarking technology to protect the content of images and track malicious users who redistribute images. Formal security analysis shows that STMIR can resist both ciphertext only attack and known background attack, and extensive experiments in the real-world image data sets demonstrate effectiveness of STMIR in terms of retrieval accuracy, retrieval efficiency, and traceability to malicious query users.