OCR is a quite well known (and old) technology and seems not to have changed much over the past decade, but its uses are of great importance in police investigations, especially when combined with the latest AI technology.
OCR, a history
Without delving into great technical detail of what OCR is, the acronym means “Optical Character Recognition”. This is, an OCR system is able to read an image (picture, photo, etc.) and to find the letters in it, reconstructing words and sentences. The output of any OCR system is always editable text.
Originally, Wikipedia credits the first OCR system to Fournier d’Albe and Tauschek, in 1870, no less. However, the first “modern” OCR systems were invented in the 1930s. This means OCR, as a technology, is about to turn 100 in a couple of years. While the most known use of OCR is as a means to convert scanned text documents back into editable text, it can be used in many other fields. And Law Enforcement is one of those fields, where OCR has an important task, aiding officers in investigations and solving crimes.
OCR in investigative police work
Besides the obvious use of scanning evidence documents in cases and converting into editable text, that can be searched, and entered into case management software (DEMS), OCR has a big role in ALPR (Automatic License Plate Recognition). In fact, any ALPR system is just a specialized OCR, that looks for number plates in an image or video frame, and converts them to text. This text can then be fed into any police database to find out if the car is involved in anything (it was stolen, was involved in a crime, has pending traffic tickets, etc.).
But it is also useful at evidence analysis. It can process screenshots (of chats, for example), printed e-mails or letters. Specialized OCR systems are able to read handwritten documents, so they can be used for ransom notes or threats.
One of the more interesting uses of OCR in Law Enforcement investigations is helping to identify locations in video recordings. Any direction or store sign can help locate the place where the video was taken. This is particularly useful in social media images, as they may contain inadvertent leads to their position.
If an OCR system is combined with automatic translation, even texts in unfamiliar languages can be used in investigations, once converted into the local language. This opens up a host of possibilities when dealing with suspects of multiple nationalities in a case, as there is no need for a translator for all of the languages.
OCR in Border control
One place where everyone has probably been in contact with an OCR system is at a border or at the passport control at an airport. The passports are scanned, images located and compared with the name that figures on the passport, to see if they match. Usually, the name is also sent out to check against international databases of wanted persons, so if anything comes up, the border agent can take the appropriate action immediately. This not only speed the border processing up, but can also be used to detect forged or altered documents. Hence it is an invaluable tool for modern border control agents.
OCR enhanced by AI
As said in the beginning of the article, OCR doesn’t seem to have changed that much, but with the enhancement of artificial intelligence capabilities, even “traditional” technologies as OCR have benefitted from it. AI provides a better pattern recognition and helps OCR identify individual words or letters. The models can also learn over time to read a specific handwriting or can be trained to identify text in bad quality images or videos, with a surprising accuracy. Thus, the combination of Deep Learning and an OCR system is quite a powerful one, as it can deal with text in any form or shape, extracting useful information from media files.
Conclusion
Even being the ugly duckling among police investigation tools, OCR is extremely useful in many investigations, especially when documents are involved (like in financial cases), but enhanced with AI, it can be applied to almost any scenario. OCR can help recognize a place by reading local number plates, signs or any text that may happen to appear on the images. It can also compile a lot of searchable information quickly, which would take an inordinate amount of hours if done manually visualizing all the material and taking notes. It helps speed up operations and avoid transcription errors.
Discover how Intelion Transforms Investigations
