Institutionen för systemteknik (ISY)

Leaf Segmentation using the Kinect

Marcus Wallenberg, Michael Felsberg, Per-Erik Forssén, Babette Dellen
SSBA11
Proceedings SSBA'11 Symposium on Image Analysis
March 2011

Abstract

Segmentation is an important preprocessing step in many applications. Purely colour-based segmentation is often problematic. For this reason, we here investigate fusion of depth and colour information, to facilitate robust segmentation of single images. We evaluate segmentation results on data collected using the Microsoft Kinect peripheral for Xbox 360, using superparamagnetic clustering. We also propose a method for aligning and encoding colour and depth information from the Kinect device. As we show in the paper, the fusion of depth and colour information produces more semantically relevant segments for scene analysis than either depth or colour separately.

Full Paper

This is a preliminary version of the work presented at the DAGM11 conference.

---

Bibtex entry

@InProceedings{wallenberg11a,
  author = 	 {Marcus Wallenberg and Michael Felsberg and Per-Erik Forss\'en and Babette Dellen},
  title = 	 {Leaf Segmentation using the Kinect},
  booktitle =    {Proceedings {SSBA'11} Symposium on Image Analysis},
  year =         {2011},
  month =        {March},
  note =         {Non-reviewed workshop},
  OPTpages =     {}
}