An image decryption technology based on machine learning in an irreversible encryption system
In this paper, a hidden image decryption system based on neural network is proposed. Hidden images converted into phase information are encoded and weighted to overlay the host image by the principles of double random phase encryption. Because the hidden image is phase information, it will be conver...
Uloženo v:
| Vydáno v: | Optics communications Ročník 541; s. 129561 |
|---|---|
| Hlavní autoři: | , |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
Elsevier B.V
15.08.2023
|
| Témata: | |
| ISSN: | 0030-4018 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Shrnutí: | In this paper, a hidden image decryption system based on neural network is proposed. Hidden images converted into phase information are encoded and weighted to overlay the host image by the principles of double random phase encryption. Because the hidden image is phase information, it will be converted into intensity image after being encrypted and received by the image sensor, which makes the system unable to retrieve the image information using the phase recovery algorithm and the conventional decryption method, so this is an irreversible coding system. However, by using a convolution neural network to process a large number of ciphertext and its corresponding plaintext, a neural network model can be trained, by which ciphertext can be directly converted to the corresponding plaintext, and the successful decryption of conventional irreversible encryption systems can be achieved. Moreover, the decryption process of the system does not need to provide the host image, so it is a blind watermarking technology. Computer simulation results show that the encryption system proposed in this paper based on machine learning decryption has high security and accuracy.
•In this paper, an irreversible coding system is proposed by watermarking and double random phase encryption method.•The system cannot be recovered by phase recovery algorithm and the conventional decryption method.•A neural network model can be trained, and the successful decryption of the irreversible encryption systems can be achieved.•Computer simulations show that the encryption system proposed in this paper based on machine learning decryption has high security and accuracy. |
|---|---|
| ISSN: | 0030-4018 |
| DOI: | 10.1016/j.optcom.2023.129561 |