code‎ > ‎

ERGC Superpixels


Eikonal-based Region Growing Clustering webpage.

Future release will contain the "moving superpixels" procedure.

Main features :
  • ERGC functions in a single .h file, so you can easily add them to your own code
  • ERGC deals with 2D/3D images with any number of channels
  • Possibility to custom the grid of seeds
  • Possibility to give your own seeds image
  • Iteratively refined version of ERGC
Brief description :
ERGC is a tool to compute easily and efficiently an over-segmentation of an image. Without any prior, the regions (or superpixels) are sampled on a grid with equal sampling step along each axis. A spatial constraint term can be used to enforce compactness of the superpixels (see eagle image above).
Note: The sampling steps can also be tuned by the user to fit morphological components of the image. Your own seed image can also be used.
ERGC is based on the CImg library. It can deal with almost any type of images, is easy to use, fast and efficient.



Installation steps :
  1. download the archive (blue arrow below)
  2. unzip 
  3. in a shell type "cmake ." then "make" to compile all the sources
See the README for usage.

ċ
ERGCSuperpixels.zip
(431k)
Pierre Buyssens,
31 oct. 2014 à 03:07