Cryptanalysis of phase-truncated Fourier-transforms-based optical cryptosystem using an untrained neural network
Wenqi He, Shuixin Pan, Meihua Liao, Dajiang Lu, Qi Xing, Xiang Peng · 2020
Optical cryptosystem based on phase-truncated-Fourier-transforms (PTFT) is one of the most interesting optical cryptographic schemes due to its unique mechanism of encryption/decryption. Conventional learning-based attack method need a large number of plaintext-ciphertext pairs to train a neural network and then predict the plaintexts from subsequent ciphertexts. In this work, we propose an alternative method of attack on PTFT-based optical asymmetric cryptosystem by using an untrained neural network. We optimize the parameters of a neural network with the help of the encryption model of PTFT-based cryptosystem, hoping to get the ability of retrieving any plaintext from the corresponding unknown ciphertext but without help of the decryption keys. The proposed untrained-neural-network-based attack approach eliminates the requirement of tens of thousands of training images and might open up a new avenue for optical cryptanalysis.