This chapter will be devoted to the ways how to hide data in photographs securely. Suppose you need to transfer an image in which you want to hide your billing information, or the person in the photo, or some other detail. In this case the so-called pixelation method is usually used, which makes a part of the image blurry, as a result of which the human eye can only recognize the content in outline.
Pixelation is often used on TV to hide the face, intimate body parts or obscene gestures. The essence of the method is very simple: the image or its part, which must be hidden, is divided into squares of the same size. The dominant color is determined in each square, and the entire square is filled with it. The larger squares are, the harder it is to distinguish the original image.
Using this method to hide valuable data in a photo is a big mistake, as modern technologies are able to recover them accurately.
Back in 2016 researchers from the University of Texas and Cornell Tech presented a program based on neural networks that restored blurry images and did not contain any innovative techniques. According to the researchers' data, they were able to recover 80% of the distorted images not only those using pixelation but also other effects such as the usual blurring.
Interview widget: Did you know about the possibility to restore the image hidden by pixelation?
But this is not the only study and not the only method to restore a distorted image, therefore, it is better to close some data with a shape, then no neural networks will help.
For example, like this: