Structured Light Based Depth Edge Detection for Object Shape Recovery

1
Structured Light Based Depth Edge Detection for
Object Shape Recovery

• Cheolhwon Kim, Jiyoung Park, Juneho Yi
• School of Information and Communication
  Engineering
• Sungkyunkwan University

Matthew Turk Computer Science Department
University of California, Santa Barbara
2
Our Approach

• Exploit distortion of light pattern in the
  structuredlight image along depth
  discontinuities
• Detect candidate locations for depth edges
  byapplying a Gabor filter

3
Problem

• Disappearance of distortion along depthedges
  depending on the distance of anobject point from
  the camera and projector

White light image
Patterned image
Amplitude map
4
Key Contributions

• Analysis and control of the parameters of
  thestructured light
• Optimal stripe width with respect to the
  distancefrom the camera/projector
• Minimum detectable depth difference
• Solutions to guarantee the occurrence of
  thedistortion along discontinuities for a
  continuous range of object location
• 1 camera 1 projector
• 1 camera 2 projectors

5
Experimental Results

• Fingerspelling Letter R

• Gesture Up

Canny edges
White light image
Depth edges
Gabor amplitude map