Optical Compressive Encryption via Deep Learning

Abstract

The compression of the ciphertext of a cryptosystem is desirable considering the dramatic increase in secure data transfer via Internet. In this paper, we propose a simple and universal scheme to compress and decompress the ciphertext of an optical cryptosystem by the aid of deep learning (DL). For compression, the ciphertext is first resized to a relatively small dimension by bilinear interpolation and thereafter condensed by the JPEG2000 standard. For decompression, a well-trained deep neural network (DNN) can be employed to perfectly recover the original ciphertext, in spite of the severe information loss suffered by the compressed file. In contrast with JPEG2000 and JPEG, our proposal can achieve a far smaller size of the compressed file (SCF) while offering comparable decompression quality. In addition, the SCF can be further reduced by compromising the quality of the recovered plaintext. It is also shown that the compression procedure can provide an additional security level, and this may offer new insight into the compressive encryption in optical cryptosystems. Both simulation and experimental results are presented to demonstrate the proposal.

Optical Correlators for Cryptosystems and Image Recognition: A Review

“…However, deep learning techniques have also been more recently proposed in an integration with optical methods, benefiting from the advantages of both [ 11 ]. Deep learning methods can address the problem of information loss in classical techniques of ciphertext compression [ 12 ]. A deep convolutional network was successfully used for repairing the compression artifact appearing in the image [ 13 ].…”

“…DL was also employed for the reconstruction of the ciphertext image that had been previously compressed into a 1D encrypted signal, using however a relatively complicated optical setup [ 15 ]. More recently, a deep neural network (DNN) was used in a novel setup, where the compression of the ciphertext was performed by a collaboration between the JPEG2000 standard and the bilinear interpolation method, whereas the decompression of the ciphertext was performed by the DNN, with results significantly better than state-of-art standards of compression, such as JPEG2000 and JPEG [ 12 ].…”