Project P12:
Shape Through Smog

Project Goal

Okay, it's not smog, But here is the story: When driving in the desert and recording hundreds of thousands of camera images, we began to notice that more distant places have a higher blue content. Here is one explanations: the Rayleigh scattering. In this project, we hope to develop techniques for acquiring range data from a single non-moving camera, by analyzing the amonut of blue in the image, possibly over extended periods of time. Here is a fashionable recent paper doing the same with fog. Check it out.

Project Scope

Develop a range finder that is based on the amount of blue in the image. This involves gathering images, coming up with an appropriate mathematical filter, modeling the physics to the extent possible, generating a range image and - most importantly - working on techniques for smoothing these range images.

Tasks

• Read papers
• generate mathematical model of blue
• collect images
• implement filter
• develop range scan
• develop range smoother

Project Status

Gregory Ng (gregng at stanford),
Ray Tsang (raytsang at stanford),
Ryan Hanson (rhanson at stanford),
Thomas Pun (dapunster at mac dot com)

Point of Contact

Adrian Kaehler and Sebastian Thrun

Midterm Report

not yet submited

Final Report

not yet submitted