OCR (Optical Character Recognition) has been around for some time now and everyone is more or less familiarized with what it does: read printed letters and numbers and convert them back into editable text. And current systems are very good at it with a 95%+ recognition rate. But handwriting is another thing entirely to be read and interpreted. However, with the latest advancements in AI it is now possible not only to read handwriting, but to identify who has written it.
AI to the rescue
Up until recently, to know if a certain handwriting corresponded to another sample and thus the same person must have written it, a trained expert was needed: a graphologist. This requires years of specialized training to be able to function in a court and to use the findings as evidence. So up until now the graphologists were only humans, doing advanced recognition work and using years of experience in their favour.
But artificial intelligence has caught up and is not only able to read printed text with predictable fonts, but also handwritten material. In fact, it has gotten so good at it, that now this technology is used for forensic analysis. This is, to compare if a certain handwriting was written by the same person than another piece of text. These systems are collectively called FISH (Forensic Information System for Handwriting) and their main goal is to analyse all the strokes, their direction, force, pressure, etc. of a text, and then compare it to another one. The main difficulty resides in that both text are usually not the same, so a simple optical comparison could be done. Hence the need for a graphologist or, now, a FISH AI.
Why use AI to compare handwritten text?
For one, using a machine can be quicker, than sending the text out to a graphologist and waiting for the results. In some police cases or investigation, time is of the essence. But the main reason is that FISH is not only a handwriting recognition system, but a database that stores all the parameters that have been analyzed in a certain text.
This way, given a new text (for example, a threatening correspondence), the examiner can scan the document into the system, and compare it with hundreds or thousands of previously stored documents, in order to find any possible matches. For a trained human it may be easy to see if two pieces of text were written by the same person, but they cannot possibly know who wrote that particular text. A FISH system can, if there is a previous sample already stored in the database. If a match is found, that information is very useful for the investigator, as it can point him or her to a specific person. And the FISH system can find matches between thousands of specimens in a matter of seconds or minutes.
Integration into a Law Enforcement case
Having the evidence that a certain letter or note was written by a certain person is useful. But normally a police case is way more complex and has many kinds of evidence, like video, audio and pictures. Our Intelion platform can integrate many different information systems, both AI and non-AI, through API calls, and shows all the relevant information for a case on one single screen, generating reports with all of it. Thus Intelion is a powerful investigative platform, capable of interfacing with multiple AI analytics tools (facial analytics, speaker ID, voiceprint, etc.), including FISH systems.
Conclusion
The analytics of handwritten text can be of great importance in certain investigations, and the (Forensic Information System for Handwriting) FISH platforms are now able to store thousands of specimens of handwritten elements, that can be compared with new material from a case, to find the author of the handwritten note or letter in a matter of minutes.