This algorithm deciphers data that you have made illegible

The year 2020 once again showed us what is possible with computer code. Intelligent algorithms develop drugs and help out in laboratories, but they can also be used for manipulation, as in deepfakes. A new tool now shows how pixelated content can be made easily readable again. The first results look promising.

Basically there can be various reasons why content is pixelated. Sensitive content on pictures or passwords on screenshots can be made unrecognizable. The method often used (linear box filter) is actually relatively simple. Only the resolution of the selected area is reduced and this is then overwritten with the average value of all pixels.

Sometimes content has to be pixelated for privacy reasons (Image: Jason Dent)

So far there have been some successes in de-pixeling content. In 2006, D. Venkatraman wrote in an article how easy it is to restore a credit card number if a pixelated image is available. All possible credit card numbers are generated, pixelated and the result is compared with the original image. However, this can hardly be used with passwords, since a lot of time would pass due to the variety of possible character combinations.

Before the newly developed algorithm can get started, it must know which font size, font color and which background will be used. Then a De Bruijn sequence (a more or less random series of characters that covers all combinations of 2 characters) with the same properties is created in the editor and a screenshot is made. This then serves as the basis for the algorithm to recognize pixelated characters.

Pixel pulp becomes readable content (Image: Sipke Mellema)

From here on, the process is similar to that for credit card numbers. The algorithm checks pixel block for pixel block and checks which letter combination could be valid here. Although there are still a few errors here and there, pixelated data could still be recovered in many cases.

This novel algorithm impressively shows the extent to which information that has been made illegible could no longer be secure in the future. Another reason why you should either cut off sensitive data completely or overwrite it. Because unlike us, the computer will always find a way to get the information it needs.

via LinkedIn


More news from the topic of security:

Leave a Reply

Your email address will not be published. Required fields are marked *