Computational photography takes a look around

Boston, MA. Cameras that look around corners, trillion-frame-per-second video, and hardware “apps” in the form of low-cost mobile-phone accessories were just a few of the topics that Ramesh Raskar addressed in a keynote presentation Wednesday at Electronics New England.

In the presentation, titled “The Capabilities of Super-High Speed Imaging,” Raskar, associate professor at the MIT Media Lab and cofounder of EyeNetra, described a historical example of high-speed imaging—Harold E. Edgerton's 1964 photo of a bullet having passed through an apple.

Today, Raskar said, researchers can achieve 1-Tfps rates, and he offered as an example a very slow-motion video of light propagating through a plastic Coca Cola bottle (you can see it here and learn more about the experimental setup).

Among other impressive high-speed imaging capabilities Raskar described was a system that can look around corners—that is, it can acquire images with no camera within the line of sight of the object being photographed. The system makes use of short laser pulses and a fast detector to resolve images through multipath scattered light. The system is described here. Applications range from cardioscopy to autonomous-vehicle navigation.

But what seems of particular importance to Raskar, and what prompted him to become involved in EyeNetra, is the prospect of bringing eyeglasses to the 2 billion people worldwide who cannot work or get an education without them. The distribution problem, he said, has largely been solved, and eyeglasses can be delivered for $3. It's the examination that presents the problem. The approach in use today uses a phoropter, an expensive piece of equipment that must be operated by a skilled and expensive professional. A Shack-Hartmann wave-front sensor offers an alternative, but it, too, is bulky and requires a trained professional.

The alternative that Raskar proposes is Netra, a low-cost cellphone attachment that can determine the required lens prescription and check for cataracts. He described it as an inverse Shack-Hartmann sensor that makes use of the phone's high-resolution programmable display to put the patient in the loop. The system combines inexpensive optical elements, and interactive GUI, and computational reconstruction to provide a complete eye exam. You can find details here.

All the applications Raskar discussed, he concluded, are based on computational photography, which involves codesign of sensors, optics, and processing capability.

Electronics New England is co-located this week with BIOMEDevice.