"Data Class-Specific Image Encryption Using Optical Diffraction"
A team of researchers at the University of California, Los Angeles (UCLA) presented a diffractive network for data class-specific transformations and optical image encryption. The team demonstrated class-specific diffractive networks that perform the desired transformations for certain input data classes. In their findings, diffractive networks were trained using deep learning and then physically fabricated using 3D printing to all-optically transform the input images and generate encrypted output patterns captured by an image sensor. The encrypted images can be decrypted only by using the correct decryption keys (i.e., the class-specific inverse transformations) to reveal the original information. The UCLA team experimentally demonstrated the proof of concept for this class-specific all-optical image encryption at near-infrared and terahertz wavelengths, validating its applicability across the electromagnetic spectrum. This class-specific encryption scheme adds an extra layer of security and makes reverse engineering of the original images belonging to the target data classes more difficult. This article continues to discuss data class-specific image encryption using a diffractive optical network.