Dave Gershgorn of the New York Times recently noted that the optics of the camera obscura have faithfully served photographers for ages. Indeed, the recipe has been kept fairly simple throughout the years: a lens, aperture, dark box and something to record the light.
“But the camera as we know it is changing,” Gershgorn confirmed. “A revolution in digital imaging research could surpass the camera obscura in almost every technical way: resolution, size and energy efficiency. It’s called computational photography, and it stems from the idea that if you can capture visual data instead of a true image, then the picture can be reconstructed with software.”
As Gershgorn reports, scientists at Rambus Labs are currently developing lensless smart sensors (LSS) based on the above-mentioned principles of computational photography.
“Instead of the traditional glass, a microscopic grating is placed over the sensor. Light spreads out as it passes through the grating, creating complex patterns on the sensor,” he explained. “The patterns are deliberately cast by the grating, so that the image can be instantly reconstructed by software. Why all the hassle? It’s for an entire camera that’s less than a millimeter thick.”
Interestingly, what would traditionally be thought of the as “the picture” is actually unrecognizable to the human eye.
“It doesn’t look like an image whatsoever, [so] we call it a blob,” Rambus Fellow Dr. David G. Stork told the publication. “It just looks like a mess, but it contains the proper information [for a wide range of applications].”
Indeed, Stork sees lensless smart sensors snapping specialized photos that no humans would ever see, such as reading QR codes to launch websites, or recognizing faces to unlock mobile phones.
According to Stork, the industry is just at the beginning of seeing what computational photography can do.
“What we capture on the sensor doesn’t have to look like an image. It’s broadening our notion of what an image is,” he added.