Project P15:
Shadow detection and removal for the DARPA Grand Challenge

Project Goal

Stanford is participating in the DARPA Grand Challenge, an autonomous robot desert race. Among others, our robot relies on camera video to detect road patches that are safe for our vehicle to traverse. A major problem so far have been shadows cast by mountains, plants and the vehicle itself. These shadow create strong edges which can make even relatively smooth road look undrivable.
The goal of this project therefore is to identify such shadowed image parts and to compensate the shadow effect.

Project Scope

Your algorithm will accept as an input a video stream and if necessary other informations provided by the race vehicle software, such as time of day or vehicle position and orientation. Your task will be to create an output video where the effect of shadow is compensated in a way that road parts in shadow will resemble parts not in shadow as closely as possible. Furthermore, the algorithm should run in real-time.

Tasks

• Detect shadow. We already have a histogram-based implementation for this which you can use, extend or replace.
• Remove shadow areas with too small image extend (easy)
• Compensate for image brightness (easy)
• Correct color hue and saturation (harder). You might need to estimate illumination conditions, such as weather (overcast, sunny, raining...) and environmental reflections.

Project Status

Chien-Yu Chen (chienyu at stanford),
Yu-shan Chang (yschang at cs),
Ping-Hsien Chin (pjchin at cs),
Wei Wei (wei.wei at cs)

Point of Contact

Hendrik Dahlkamp

Midterm Report

not yet submited

Final Report

not yet submitted