Korea Times quoted Park Sang-jin, president of Samsung’s digital imaging division, saying "As the issue with 3D televisions is providing a glass-free viewer experience, 3D cameras has a similar challenge for achieving a one-lens, one-imaging sensor approach. The two-lens, two-sensor 3D camera released by Fuji is still too expensive and inconvenient for users."
Park also said that the company may produce a camera capable of taking three-dimensional (3D) images sometime next year, but admitted that it will be a digital guinea pig, saying that the "real" 3D cameras that are suited for conventional use won’t probably be available until after 2012.
I wonder what is the measurement principle behind Samsung's single-lense single-sensor 3D camera. Time of flight or structured light?
ReplyDeletesingle lens, single sensor, but a 1-W laser to burn your eyes !
ReplyDeleteI believe Park is talking about stereo imaging, as the human vision is essentially stereo, rather than 3D. 3D term is often used for both stereo and depth measuring cameras, and this is misleading. For example, the Fuji camera he mentions mimics human stereo vision with its two-sensor and lens approach.
ReplyDeleteStereo imaging with just one lens and one sensor? Is that possible?
ReplyDelete@ Stereo imaging with just one lens and one sensor? Is that possible?
ReplyDeleteYes, there are ways to do it.
The only way I can think of is Coded aperture along the line of what Ramesh Raskar does at MIT like the reinterpretable imager:
ReplyDeletehttp://nuit-blanche.blogspot.com/2010/04/cs-reinterpretable-imager-towards.html
Igor.
Wow. I guess my secret project is not a secret anymore. One lens, one sensor, yes it is possible but the issue is of course performance level. Stay tuned.
ReplyDelete@ The only way I can think of is Coded aperture
ReplyDeleteAnother possibility is described in Kodak patent application US20090219432. There are some other ways too.
FYI, :-)
ReplyDeletehttp://nuit-blanche.blogspot.com/2010/09/cs-i-dont-really-care-what-you-call-it.html
(Nuit Blanche is a blog mostly focused on the topic of Compressed Sensing / Compressive Sensing)
Igor.
At Stanford, Fife and El Gamal described 3D through single objective lens using very small pixels. Some of this was at ISSCC a few years ago.
ReplyDeleteSee www.visionsense.com for a miniature single stereo camera design that provides high image quality; is Proven & used in the medical field for years.
ReplyDeleteAll these single lens stuffs can give a 3D "perception" for a human observer. The principle is well known, but for 3D depth measurement, the precision that you can obtain is very low. In fact, the "3D" perception comes mainly from human observer's experience not realy from measurable 3D depth.
ReplyDeleteStereo on a chip is very doable. The resolution is there and also some of the algorithms already support applications.
ReplyDeleteRe: "3D" perception comes mainly from human observer's experience not realy from measurable 3D depth
It's actually a combination of both. 3D perception is both genetic as well as developed. 3D depth sensing is achievable with disparity computation: pixel-based, edge-based, motion-based, etc...