Pixelating images to hide details can be reversed with a free app

Unredacter can frustrate attempts to hide text using pixelation or blur effects, rebuilding the original text with minimal effort.

If in the case of marking the text with printed or printed documents, the procedure can theoretically be reversed by using sophisticated means, mastered only by forensic scientists and professionals in the field of intelligence agencies, the equivalent applicable to scanned and digitally stored documents should be infallible, only it is not.

Except for completely opaque color coverage, hiding text from scanned documents using digital blur or pixelation effects can be surprisingly easy to reverse. It all starts with the fact that these pixelation filters use mathematical algorithms to determine the level of gray applied to each square of the hidden text, the calculation starting by analyzing the source material can be reversed to determine with reasonable accuracy each letter of the original text.

If in the case of a pixelated picture depicting a person’s face, the reconstruction of the original image involves a lot of effort of “guessing” from an artificial intelligence system helped by the “experience” of millions of examples detailing all aspects of human physiognomy, pixelated pictures that only cover text elements can be rebuilt much easier. Starting with the uncovered portions of the text, determining the character set, font size and type is a trivial matter, leaving only the letter-by-letter extrapolation of the original text by matching the distorted fingerprints by applying the pixelation effect.

Theoretically, the text reconstruction process can also work with documents to which blur effects have been applied, if the intensity used is low enough to leave clues about the shape of the original text.

How can you avoid disclosing sensitive information by reversing pixelation or blur effects? Very simple. You need to mark text portions using only solid colors, which leave no room for analysis and reconstruction of the original information.