| Autor: |
Deeba, Farah, Kun, She, Dharejo, Fayaz Ali, Memon, Hira |
| Předmět: |
|
| Zdroj: |
Information Security Journal: A Global Perspective; 2020, Vol. 29 Issue 1, p30-39, 10p |
| Abstrakt: |
The tremendous AI benefits, machine learning and deep learning, have led to the adoption of advanced technology and modern applications. Especially in an image, Video processing, natural language processing, speech recognition. AI algorithms have recently overcome many drawbacks, thanks to the DNN models, that have contributed to delivering state-of-the-art results in computing and other areas. But safety and security are always challenging tasks.Our proposed approach provides a secure and efficient watermarking method based on a neural network for digital images using the least significant method. First, we used the least significant bit (LSB) to insert a watermark for the image pixel. Because only LSB -based methods are not robust; they are not sufficient in an attack-free environment and lossless compression. We used an Artificial Application Neural Network (ANN) to detect the presence of sensitive information and extract information from the source image. It is inherently unstable when the proper machine learning algorithm is trained, re-trained, and adapted to a few new applications. The standard solution would have a digital signature there as there are very simple ways to change the neural network model so that it still does the same thing as before, but the overall presentation will be different. This paper highlights the essential needs of the ANN model in watermarking. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
|
Nepřihlášeným uživatelům se plný text nezobrazuje |
K zobrazení výsledku je třeba se přihlásit.
|
