Hi all, i have read some literaure on light fiield depth estimation and it seems most of them use either resolution information or shading as a parameter to obtain or estimate depth of the scene. Are there other methods that are accurate or that could be used to estimate depth in a scene?
Depth is a very funny word in English and can have unexpected usages and meanings. Assuming that you mean distance between the eye and an object in the field of view (scene), of course the most straightforward way camera can o it is by measuring the time that light travels to and fro a piece of a scene, just like with a RADAR. This is called 3D LADAR, google it.
Yes,
You can use instead of depth of a scene by an optical path difference (OPD). Then you well find a lot of optical methods which cen be used to measure OPD such as interferometry and digital holography.
Hope this helps
Thanks Mario and Mostafa.
@Mario, I mean depth in the scene e.g. between foreground and background.
@Mostafa, I didn't get the exact method you are proposing, is there something missing in your answer?
Edgar,
It is hard to elaborate such a wide topic efficiently over this channel. Please have a look at some methods mentioned above, in particular OCD, OCT, holography and 3D LADAR, and get back if you have a precise question. Personally I am most impressed by 3D LADAR, for example see the presentation of Princeton Lightwave and a small video:
http://www.princetonlightwave.com/mm-products/3d-ladar
Dear Edgar Owiti,
the fundamental information required is the offset between image parts due to shifts in perspective. All light field cameras are implicitly based on this principle: Depending on the perspective, objects in the background appear offset (to the side) from the objects in the foreground. A light field camera basically records multiple perspectives. By projecting these different perspectives onto each other, you get an image that is sharp in the foreground and blurry in the background, or vice versa. Based on this information, you can also calculate the light field, that is, a three-dimensional set of light rays. Please read "Seeing lens imaging as a superposition of multiple views" on arXiv.
Kind regards,
Sascha Grusche
Hi Edgar,
Light field imaging is not my primary field, but there is a paper describing a method to extract the 3D location from a light field using a 4D phase space. The paper states: "The concepts presented here, however, are general and can be used with any phase-space dataset (e.g. light field cameras)."
"The key idea in this paper is that each point source in 3D traces out a unique 2D plane within the 4D phase space." Maybe that could answer your question about the fundamental information related to depth/location in the light field. In the paper they are able to find the 3D location of point sources. I don't know if it can be extended to bigger objects.
https://www.osapublishing.org/oe/abstract.cfm?uri=oe-23-11-14461
I hope it helps.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Optics