This website requires JavaScript.

Data class-specific all-optical transformations and encryption

Bijie BaiHeming WeiXilin YangDeniz MenguAydogan Ozcan
Dec 2022
Diffractive optical networks provide rich opportunities for visual computingtasks since the spatial information of a scene can be directly accessed by adiffractive processor without requiring any digital pre-processing steps. Herewe present data class-specific transformations all-optically performed betweenthe input and output fields-of-view (FOVs) of a diffractive network. The visualinformation of the objects is encoded into the amplitude (A), phase (P), orintensity (I) of the optical field at the input, which is all-opticallyprocessed by a data class-specific diffractive network. At the output, an imagesensor-array directly measures the transformed patterns, all-opticallyencrypted using the transformation matrices pre-assigned to different dataclasses, i.e., a separate matrix for each data class. The original input imagescan be recovered by applying the correct decryption key (the inversetransformation) corresponding to the matching data class, while applying anyother key will lead to loss of information. The class-specificity of theseall-optical diffractive transformations creates opportunities where differentkeys can be distributed to different users; each user can only decode theacquired images of only one data class, serving multiple users in anall-optically encrypted manner. We numerically demonstrated all-opticalclass-specific transformations covering A-->A, I-->I, and P-->I transformationsusing various image datasets. We also experimentally validated the feasibilityof this framework by fabricating a class-specific I-->I transformationdiffractive network using two-photon polymerization and successfully tested itat 1550 nm wavelength. Data class-specific all-optical transformations providea fast and energy-efficient method for image and data encryption, enhancingdata security and privacy.