Cloud is safe when compressive: Efficient image privacy protection via shuffling enabled compressive sensing

Abstract

Cloud-assisted image services are widely used for various applications. Due to the high computational complexity of existing image encryption techniques, privacy protection becomes extremely challenging for resource-constrained smart devices. We propose eCIS, a cloud-assisted image service where compression and encryption are jointly used for image processing efficiency. eCIS could shift the high computational cost from encoder and decoder to the cloud while providing efficient and adaptive image privacy protection for end users. The key idea is to leverage different measurement matrices for sampling device and cloud. Thus encryption is realized with shuffling enabled compressive sensing. We conduct in-depth theoretical analysis and demonstrate the effectiveness with evaluations. To this end, extensive experimental results show that eCIS can effectively protect image privacy and meet user’s adaptive security requirements. Meanwhile, our experimental results show that eCIS could significantly save the system running time compared with existing cloud-assisted schemes.