Schoolchildren all over the world know the way to write secret messages on paper using lemon juice or wax or other household substances. The invisible message is written with the appropriate material and is only revealed when the recipient “decodes” it using heat or some other way to develop the hidden substance.
A more sophisticated approach to secret messages is needed in the adult world, of course, and there are many different tools that allow sensitive documents to be encrypted beyond brute-force attack so that only the legitimate recipient can read them. Such technology works optimally with the digital output of word-processing and related software where the bits and bytes of the document can quickly and efficiently be scrambled using a password or key. The reverse process is then only available to the holder of the key.
However, there is a problem when it comes to handwritten documents. A scanned image of such a document is not composed of bytes representing the letters and words of the document, rather it is a map of all of the pixels making up the document. As such, a handwritten document might be encrypted by applying an appropriate tool for image encryption providing the scanned document is of sufficiently high resolution. Either way, there will be a lot of redundancy in the encrypted image file. This means greater processing power is needed for the initial encryption, the encrypted document file size will be larger than necessary, and the decryption process itself will use excessive processing power to retrieve the original document.
Such matters are perhaps of little consequence when considering a short segment of handwriting, but a handwritten report running to many pages would best be encrypted with a more efficient technology aimed specifically at the written word.
Writing in the International Journal of Information and Computer Security, a team from the Higher Colleges of Technology in Abu Dhabi, United Arab Emirates, have demonstrated such a technology in the form of a handwriting document encryption scheme based on segmentation and chaotic one-dimensional logarithmic map. The approach takes the scanned document and breaks it up the words digitally into their component parts, grapheme. The pixel locations of each part of the grapheme rather than the whole scanned area of the document are then scrambled with the encryption key. The team has offered proof of principle with standardized test documents and demonstrates how efficient their process is.
The team explains that there are 2 to the power of 180 (2180) possible encryption keys for their approach, which makes it immune to brute-force attacks with current computers. Moreover, their statistical analysis indicates superior permutation and substitution properties for their proposed encryption scheme compared with conventional image encryption schemes applied to the same test documents. The process is relatively slow but the team is now optimizing performance for real-world applications. One additional benefit is that the same technology might also be adapted to different alphabets and perhaps even character-based languages without compromising the performance and efficacy.
Abu-Amara, F. and Bensefia, A. (2021) ‘A handwriting document encryption scheme based on segmentation and chaotic logarithmic map’, Int. J. Information and Computer Security, Vol. 14, Nos. 3/4, pp.327–343.