Steganography: when one content hides another

Less known than cryptography, steganography is the other way we have to transmit a message confidentially, hiding it in other content. Explanations.

Everyone knows the technique of the message written in lemon juice: invisible to everyone’s eyes but which is revealed when the paper is placed near a heat source. This process, which consists of concealing one’s message in a so-called “cover” medium, is called steganography, from ” steganos meaning “tight” in Greek. It has long been based on ingenious but rudimentary processes, which have been replaced by digital techniques which have made many steganographic tools available to the general public (PixelKnot, Steghide, Invisible Secret 4, etc.). Digitization now makes it possible to hide any message in another audio, written or pictorial content.

While cryptography leaves the message visible but makes it unreadable to anyone who does not have the decryption key, steganography consists of making this message almost invisible to everyone. Cryptography is sufficient to exchange information that we just want to keep confidential, such as bank data; steganography is necessary for those who cannot allow themselves to be suspected that they are hiding something, which is for example the case of dissidents living in totalitarian regimes, spies… or criminals.

Hide a needle in a pixel boot

The use of steganography grew considerably at the time of the September 11 attacks. Al-Qaeda agents were then suspected of using the methods of steganography to convey messagesnotes Patrick Bas, director of research at the Center for Computer Science, Signal and Automation Research in Lille. In response, and especially later, in the context of investigations into pedophile networks, security agencies had to develop new decryption techniques. “. In this case, we speak of steganalysis.

Sympathetic ink revealed under ultraviolet light

While there are several types of digital concealment – ​​in sounds, videos, and even other texts – the currently most widespread steganographic method is based on images. We speak in this case of spatial insertion: the steganographic tool selects the positions of the most “discreet” pixels to hide a message there. ” In a portrait, for exampleexplains Patrick Bas. The pixels that form a person’s hair are the most complex areas of the image and offer better hiding places than a homogeneous area like a piece of wall in the photo where the slightest modification can be quickly spotted.…” The goal here is to determine the areas of the graphic content that are the most difficult to analyze.

Moreover, a digital image of standard size is made up of a million pixels which each contain a color value coded on several bits: we thus have several million bits in which our secret message can be hidden. The steganographic method will consist of modifying some of these bits, coding for information that is difficult to detect with the naked eye (such as the level of saturation). They are called “weak point bits”. All these micro-modifications will then be located through the use of the encrypted key.

Steganalysis or the AI ​​version of Sherlock Holmes

Researchers have naturally developed methods to try to locate and decrypt hidden messages. This is how shorthand gave birth to steganalysis. These methods are primarily based on software deep learning (deep learning) that inspect thousands of pieces of content, with the aim of gathering common characteristics and differences among images that contain messages and images that do not.

Steganography when one content hides another

On the left: image containing a hidden message (the insertion of the message is invisible to the eye), on the right: amplified difference between the original image and the image containing hidden information: this method modifies the image on the pixels which are the least detectable.

Steganalysis also consists of detecting the presence of anomalies or irregularities referred to as noise. A “natural” noise that we find in all our photos, to which is added the “artificial” noise introduced by the steganographic tool. A photo taken in broad daylight will thus be much sharper and more faithful to the object photographed than a photo taken in the middle of the night, where the photons of the captured light are few and invaded by the “noise” of dark and coarser pixels. which interfere with the image. It is by analyzing this photonic noise, ie counting the number of photons, that the stegananalyst will attempt to detect all anomalies and extract the steganographed message.

This is why hair, a very rich and complex area where there are many fluctuations, will be much more conducive to hiding a message. Since there is already the presence of many parasites or photonic noise, the micro-modifications will have a better chance of going unnoticed! concludes Patrick Bas.

We want to say thanks to the author of this short article for this incredible material

Steganography: when one content hides another

Take a look at our social media profiles as well as other pages related to them