Privacy-Preserving Image Classification With Deep Learning and Double Random Phase Encoding

With the emergence of cloud computing, large amounts private data are stored and processed in cloud. On other hand, owners (users) may not want to reveal information providers protect their privacy. Therefore, users upload encrypted or third-party platforms, such as Google Cloud, Amazon Web Service, Microsoft Azure. Conventionally, must be decrypted before being analyzed cloud, which raises privacy concerns. Moreover, decryption big images requires enormous computation resources, is unsuitable for energy-constrained devices, particularly Internet Things (IoT) devices. Data most popular applications, image query classification, can preserved if directly classified on IoT devices without decryption. This paper proposes a high-speed double random phase encoding (DRPE) technique encrypting into white-noise images. DRPE-encrypted then uploaded Images that using deep convolutional neural networks The simulation results indicated feasibility good performance proposed approach. privacy-preserving classification method useful data-sensitive fields, medicine transportation.