How to detect how far the object on photo is from another objects on that photo?

Question

So for example we have real life photo. how to get (relativly to image dimentions for example) the distance from wall to girls, from girls to trees if all we know ts this picture?

_{(source: vkontakte.ru)}

Papers with algorithms and\or Open Source programs doing this would be appreciated. As for me Code is Better than just pure algorithm. So if you give algorithms please give links to Open Source programatic representations of that algorithms!)

So How to detect how far the object on photo is from another objects on that photo?

Answer 1

In order to measure the distance of objects from a photo, you need to map these objects in a space representation. If you expect real distance like in real world (3D), you need a 3D space representation.

First, it is better to look at lens and spatial distortions. http://www.beyondmegapixels.com/2008/12/spatial-relationships-and-distortion/

Back to the space representation, you need to understand how to build a 3D scene using photography. http://www.feargod.net/3dhowto.php

Here is a paper talking about this topic. http://www.isprs.org/education/PDF/DAGM-Tutorial-Color.pdf

well, i know the microsoft photosynth ( http://photosynth.net) that can do the reconstruction. ref: http://on10.net/blogs/larry/Fotomo-3D-scenes-from-photos/

Here it is a interesting example of how this technology works http://olympus.co.jp/en/magazine/pursuit/feature_article/…

Googling for photosynth open source projects:
http://code.google.com/p/pixelstruct
http://phototour.cs.washington.edu/bundler
http://openphotovr.org
http://photosynth.net/discussion.aspx?cat=6b63cb81-8b57-4d5d-a978-41d5509bf59a&dis=1de0ba11-2eea-4cb3-907a-252fb30f1200 (more link)