MIT's New Imaging Method Could Let You Read Closed Books

Novel X-Ray Vision

We have all sorts of sensors and gadgets that can analyze objects right in front of them. There's virtual reality tech that can read eye movements, artificial intelligence that can understand movies, and algorithms that scan and interpret text.

But what if these machines could see through objects? That is exactly what researchers at MIT and Georgia Tech have developed, a prototype system that can read books while they are still closed. The system scans books using radiation and interprets the return signals, telling you exactly the letters and words hidden behind a book cover.

 

Well, kind of. Currently, the algorithm can only interpret the distance from the camera to the top 20 pages in a stack, but past a depth of nine pages, the energy of the reflected signal is so low that the differences between frequency signatures are disrupted by noise.

Despite these limitations. People are already excited.

“The Metropolitan Museum in New York showed a lot of interest in this, because they want to, for example, look into some antique books that they don’t even want to touch,” says Barmak Heshmat, MIT researcher.

“It’s actually kind of scary,” adds Heshmat. “A lot of websites have these letter certifications [captchas] to make sure you’re not a robot, and this algorithm can get through a lot of them.”

Radiation scanning

The paper that documents the research, published in Nature Communications, describes MIT's system. The system uses terahertz radiation, the band of electromagnetic radiation between microwaves and infrared light. A terahertz camera emits ultrashort bursts of radiation aimed at the target, and the signals that return are analyzed to determined which bounced at ink, and which interacted with blank paper.

This data is then fed to an algorithm that produces an image of the page. According to Popular Mechanics, The original MIT algorithm produces the raw image from the sensor. A Georgia Tech algorithm then mucks through the noise generated, signals that have bounced too much to be of use or the background hum of the machines, and determines the individual letters.

The researchers are now working at improving the system, allowing it to accommodate more pages and reduce the "noise" it detects. Eventually the researchers hope to develop a system that can see through entire books — right past the cover.