Image Identification of Grayscale-Based JPEG Images for Privacy-Preserving Photo Sharing Services

We propose an image identification scheme for double-compressed JPEG images encrypted by a grayscale-based encryption method, where the encrypted JPEG images are referred to as grayscale-based JPEG images, that has been proposed for Encryption-then-Compression (EtC) systems with JPEG compression. The proposed scheme aims to identify encrypted JPEG images that are generated from an original JPEG image. To store images without any visual sensitive information on photo sharing services, grayscale-based JPEG images are generated by using the grayscale-based encryption method. The use of the grayscale-based JPEG images and feature vectors extracted from the JPEG images allows us to identify images not only recompressed multiple times but re-encrypted with different keys. In an experiment, the proposed scheme is shown to have a high identification performance, even when images are recompressed multiple times and re-encrypted with different keys.