Unsupervised RGB-D image segmentation using joint clustering and region merging

Abstract : Recent advances in imaging sensors, such as Kinect, provide access to the synchronized depth with color, called RGB-D image. In this paper, we propose an unsupervised method for indoor RGB-D image segmentation and analysis. We consider a statistical image generation model based on the color and geometry of the scene. Our method consists of a joint color-spatial-axial clustering method followed by a statistical planar region merging method. We evaluate our method on the NYU depth database V2 (NYUD2) and compare with existing unsupervised RGB-D segmentation methods. Results show that our method is comparable with the state of the art methods. Moreover, it opens interesting perspectives for fusing color and geometry in an unsupervised manner.
Complete list of metadatas

https://hal-ujm.archives-ouvertes.fr/ujm-01020565
Contributor : Olivier Alata <>
Submitted on : Tuesday, July 8, 2014 - 11:38:23 AM
Last modification on : Wednesday, July 25, 2018 - 2:05:31 PM

Identifiers

  • HAL Id : ujm-01020565, version 1

Collections

Citation

Abul Hasnat, Olivier Alata, Alain Trémeau. Unsupervised RGB-D image segmentation using joint clustering and region merging. British Machine Vision Conference (BMVC), 2014, Sep 2014, United Kingdom. pp.1-12. ⟨ujm-01020565⟩

Share

Metrics

Record views

380