Reconstructing an image out of a white surface light information.

[Ovden151]

After watching these videos about Computational photography:

and lensless image sensors:


was wondering if it is at least theoretically possible to calculate the light of the surroundings of a white surface just from the light information on that surface.

For instance taking a picture of a rectangular white paper and then compute how all the things in front of it looked like just by extracting the light information on the white paper.

Are there any algorithms or papers about this?

 

[Ovden151]

Sorry for the bump... If this subject needs a better explanation just let me know.

 

[Drakkith]

Your question is a little unclear. What does "calculate the light of the surroundings" and "compute how all the things in front of it looked" mean? Are you trying to use the white paper as a type of "mirror"?

 

[Khashishi]

Those videos are too long to expect us to watch before answering your question. So you need to ask the question in a way that we will understand without watching them.

 

[Ovden151]

Thanks for the heads up guys. Here is a graphic describing the idea, and a short video of what's described in the second video above.

The first video of the main post is about computational photography and shows some examples of techniques from that field.
For instance it shows how computationally it can be separated the direct light and the global illumination affecting different objects. They came up with algorithms that can separate bounced indirect light, sub-surface scattered light, from all the direct light. They also show how they could remove the motion blur of all moving objects no matter the speed difference and direction of each object.

And the second shows a lensless image sensor which gathers all the light around, it doesn't generate the final image, but later the image is computed analyzing the light gathered by the sensor. It seems to gather light from all directions, and with the light information saved, it can even allow to chose the focusing of the image *after* capturing the light information.

 

[Ovden151]

In this video they show a way to infer the image from a point of view different than the one of the camera.

 

[Drakkith]

For instance taking a picture of a rectangular white paper and then compute how all the things in front of it looked like just by extracting the light information on the white paper.

If I understand the last video correctly, no. It seems to me that they illuminated one section of the playing card at a time using the projector. Without the ability to control the illumination you can't get a clear image from the white surface since light from the entire card is mixed together.