[gimp plugin user manual]

Quick Start:

  1. After installation, the plugin can be found in GIMP's menus under Filters/Misc/Foreground Extraction....

  2. Filter/Misc/Foreground Extraction

  3. Select region of interest
    The region of interest must contain the entire object you want to extract and as few background as possible. Drag a box around the object with your mouse. Once you have finished making your selection click the Apply Selection

  4. Select Superset of Foreground

  5. Select some known foreground
    For better results you must specify one or more representative subsets of the foreground object. These regions must not contain any background pixels. Once you have made a selection click Apply Selection to add it. In most cases one or two should be enough, use more for really tricky images.

  6. Select Subsets of Foreground

  7. Start object extraction
    Now click the OK button to start separation. Depending on your hardware and the image size, the result will be given in a few seconds.

See the Result

Advanced Options:

If you are not satisfied with the result, you can fine-tune some parameters:
  • Use the smoothness factor to fine-tune border accuracy. If your image contains a lot of undesired "holes" use more smoothing. If you find the boundary isn't accurate enough use less smoothing.
  • Sometimes you can help the algorithm by using GIMP's freehand selection tool before starting the plugin.
  • If your image contains a lot pixels of similar color (for example, everything is in different yellow tones) you can use the one ore more of the L/a/b sliders to increase or decrease the resolution in each of the color bands.
  • Consult the FAQ page.
If you are still not satisfied: Please note that this is only a demonstrator. The actual tool in GIMP 2.4 will be a lot more effective due to better interaction and preview possibilities. However, there are still a lot of images that cannot be segmentated because of various problems (for example, a true segmentation might just not exist!).

The horse picture can be downloaded at www.bigfoto.com.

Copyright 2005, 2006 by SIOX Team. All rights reserved.