pictures hide images
Technology Research News
It's not hard to hide information in a
digital image, but when the image is printed out, the hidden information
is usually left behind on the computer.
A pair of researchers from Ben Gurion University in Israel have come up
with a way to hide an image within a printed picture, allowing hard copies
of an image to retain hidden information.
The scheme hides one half-tone picture in another so that scanners like
those in supermarkets can unlock and view the information. The technique
could be used to hide barcodes in product labels and fingerprints in ID
The researchers’ system puts together two data files for two images, “one
that we want to print as an observable picture and the other that we want
to conceal within the observable picture," said Joseph Rosen, associate
professor of electrical and computer engineering at Ben Gurion University.
The second, hidden image is coded into the first picture. The researchers
scramble the mathematical representation of the hidden image with a mathematical
key. Once an image is encoded, "only an authorized person who has the
key... can reveal the hidden image,” Rosen said.
The composite image can be printed on any printer; the print can then
be read by a conventional optical scanner and processed by a PC to access
the hidden image, said Rosen.
The method works because almost every printed picture is a halftone image,
meaning it is actually a series of discrete dots. But because these dots
are too small for the naked eye to see, a black and white picture appears
many shades of gray. “A gray-tone printed picture is a collection of many
white-only dots on a black background. The size of the dots, or their
density, defines the level of the gray-tone that we see,” said Rosen.
The hidden image is encoded in the position of the visible image's dots.
Each dot can be shifted without significantly changing the visible image,
and the altered positions can be mapped as a mathematical code.
Although the researchers have only hidden gray-tone pictures, the same
method could be used for color, he said. A color picture is “still a composite
of dots but this time there are 3 color dots -- red, green and blue,”
While the visible image looks slightly fuzzy, there is no way for the
eye to discern the hidden image. The deciphering depends on the key. “The
correlation is an operation of scrambling the key... with one image in
order to get the other image,” he said.
This system is optical and different from digital watermarking, which
hides information within the numbers that make up the digital image, said
Rosen. “Instead of having digits to play with, we have binary dots that
we manipulate,” Rosen said.
Printed copies of the researchers' pictures still contain the hidden file,
but printouts of digital files that contain watermarks do not retain those
watermarks. "When you print out a [digital] file the numbers become gray
levels and the hidden information gets lost,” he said.
Like a hologram, the hidden image is concealed in a global manner, said
Rosen. “Every part of the [visible image] contains information on the
entire hidden image, such that if you cover or destroy part of the concealogram
you can still recover the entire hidden image from the rest,” he said.
The hidden image can be elicited even when 55 percent of the halftone
picture is damaged or missing, he said.
The method can be used to conceal any image, including barcodes and fingerprints.
College students, for instance, could have their meal-plan barcodes embedded
in their ID photos; the labels of prescription medicine bottles could
include information for the pharmacist about counter-indications, compliance,
and pricing in one easily scanned label, said Rosen.
The work looks good, said Chris Honsinger, a senior research scientist
at Kodak. While it seems similar to digital watermarking, this work is
the “first real optical implementation I have seen. Most watermarking
techniques require analog to digital conversion so that they can perform
operations only possible using digital computation,” he said.
The technique could be in practical use in a year, Rosen said. The researchers
next plan to improve resolution of the images. “With a better PC, printer,
scanner and a software engineer devoted to the project, we can reach a
much better quality,” said Rosen.
Rosen's research colleague was Bahram Javidi of the University of Connecticut.
They published the research in the July 10, 2001 issue of the journal
Applied Optics. The research was funded by the Ben Gurion University.
Timeline: <1 year
TRN Categories: Cryptography and Security; Computer Vision
and Image Processing
Story Type: News
Related Elements: Technical paper, "Hidden Images in Halftone
Pictures," Applied Optics, July 10, 2001
Printed pictures hide
Quantum demo does
Email burdened by
Pixel feedback forms
Software agents evolve
Research News Roundup
Research Watch blog
View from the High Ground Q&A
How It Works
News | Blog
Buy an ad link