ML-based steganalysis

One of the main techniques in steganography is hiding messages in images by altering the least significant bits (LSB) of the pixels with those of the message bits. The result is an image with a message hidden in it that the human eye cannot distinguish from the original image. This is because, on changing the LSB in the pixels of an image, the pixel values are only altered by a small amount, resulting in a visually similar image.

There are two prominent methods for LSB:

  • The naïve method is called LSB replacement. In this method, the LSB bit remains unchanged if the message bit is the same as the LSB; otherwise, the bit is altered. Hence, the odd pixels are reduced by 1 in intensity, whereas the even pixel values are ...

Get Machine Learning for Cybersecurity Cookbook now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.