A contrario segmentation: Acsegmentor

Contact: nicolas.burrus@gmail.com

Contact: Nicolas.Burrus@uc3m.es

• Source acsegmentor-0.2.0a-Source.zip

• Version 0.2a (18 January 2009)

• Changes

• NEWS
  • GUI improvements.
  • Better default parameters.
  • Compilation fixes for newer gcc (v4.3.2).

More details can be found in this paper PDF to appear in Pattern Recognition or in this paper (in french) presented to RFIA 2008 PDF.

• Version 0.2a (18 Janvier 2008)
  • Source acsegmentor-0.2.0a-Source.tar.bz2 or acsegmentor-0.2.0a-Source.zip (bigger)

• amd64: acsegmentor-0.2.0a-Linux-amd64.tar.bz2

• Other platforms
  • No binaries yet. Compilation has not yet been tested on Windows neither Mac OS X.

• Version 0.1 (14 September 2007)
  • Source acsegmentor-0.1.0-Source.tar.bz2 or acsegmentor-0.1.0-Source.zip (bigger)
  • Linux binaries

     exploration heuristic and egbis initialization with sigma=0.8 K=150 minsize=20

  $ acsegmentor —image lena.pgm —output myoutput.pgm —initializer watershed
  => idem as above with a watershed initialization with default parameters
     (sigma=0.8), result is in `myoutput.pgm’

  $ acsegmentor —learn —threshold-database thresholds.xml                 —width 1024 —height 1024 \

More details can be found in this draft paper PDF submitted to Pattern Recognition or in this paper (in french) presented to RFIA 2008 PDF.

(:title ''A contrario'' segmentation: Acsegmentor:)

(:title ''A contrario'' segmentor: Acsegmentor:)

(:title A contrario segmentor: Acsegmentor:)

(:title A contrario segmentor:)

Contact: Nicolas.Burrus@ensta.fr

• Intel (tested on Tiger 10.4.10): acsegmentor-0.1.0-Darwin.zip

• Intel (tested on Tiger 10.4.10): acsegmentor-0.1.0-Darwin.tar.bz2

• Intel (tested on Tiger 10.4.10): acsegmentor-0.1.0-Darwin.tar.bz2

• amd64: acsegmentor-0.1.0-Linux-amd64.tar.bz2

• Source acsegmentor-0.1.0-Source.tar.bz2 or acsegmentor-0.1.0-Source.zip (bigger)

• x86: acsegmentor-0.1.0-Linux-x86.tar.bz2
• amd64: acsegmentor-0.1.0-Linux-amd64.tar.bz2

• Intel (tested on Tiger 10.4.10): acsegmentor-0.1.0-MacOSX11-intel.tar.bz2

• Windows binaries (32bits, tested with VC8 and mingw with WinXP): acsegmentor-0.1.0-Windows.zip

• Intel (tested on Tiger 10.4.10): acsegmentor-0.1-MacOSX11-intel.tar.bz2

• x86: acsegmentor-0.1-Linux-x86.tar.bz2
• amd64: acsegmentor-0.1-Linux-amd64.tar.bz2

• Intel (tested on Tiger 10.4.10): acsegmentor-0.1-MacOSX11-intel.tar.bz2

• Windows binaries (32bits, tested with VC8 and mingw with WinXP): acsegmentor-0.1-Windows.zip

• x86 acsegmentor-0.1-Linux-x86.tar.bz2
• amd64 acsegmentor-0.1-Linux-amd64.tar.bz2

• Intel (tested on Tiger 10.4.10) acsegmentor-0.1-MacOSX11-intel.tar.bz2

• Windows binaries (32bits, tested with VC8 and mingw in Win XP) acsegmentor-0.1-Windows.zip

• Mac OS X binaries

• MacOSX binaries

• Linux binaries

• `MacOSX binaries
  • Intel (tested on Tiger 10.4.10): acsegmentor-0.1-MacOSX11-intel.tar.bz2
• Windows binaries (32bits, tested with VC8 and mingw in Win XP)acsegmentor-0.1-Windows.zip

• Version 0.1 (14 September 2007)

More details can be found in this draft paper PDF submitted to Pattern Recognition.

More details can be found in this ***draft paper*** PDF submitted to Pattern Recognition.

   for the GUI version, or type acsegmentor --help to find the (numerous) available
   command line options.

Acsegmentor uses CMake (http://www.cmake.org) to generate Makefiles, KDevelop, XCode or VisualStudio projects. Once cmake is installed, you just need to run it on the root directory of the source tree. It is available in most Linux distributions.

• Intel: acsegmentor-0.1-MacOSX11-intel.tar.bz2

• There are a console mode and a GUI mode. Use it without command line parameters

for the GUI version, or type acsegmentor --help to find the (numerous) available command line options.

  $ acsegmentor —image lena.pgm

  $ acsegmentor —image lena.pgm —output myoutput.pgm —initializer egbis

     (k=300, sigma=0.8), result is in `myoutput.pgm’

  $ acsegmentor —image bigimage.pgm

  => Thresholds were pre-computed for some 256×256 images. 
     For other sizes, you have to learn new thresholds
     (see LEARNING EXAMPLES).

  => 1000 noise images will be analyzed to deduce statistical thresholds.

• Note: The code still has a “research flavor”, i.e. suffer from incremental development design issues.

• Use acsegmentor --help to find the (numerous) available command line options

• Version 0.1 (5 September 2007)

Acsegmentor uses C Make (http://www.cmake.org) to generate Makefiles, KDevelop, XCode or VisualStudio projects. Once cmake is installed, you just need to run it on the root directory of the source tree. It is available in most Linux distributions.

• MacOSX binaries

• Windows binaries, run it with command.com or Msys to provide command line arguments: acsegmentor-0.1-Windows.zip

• Version 0.1
  • Source acsegmentor-0.1-Source.tar.bz2 or acsegmentor-0.1-Source.zip (bigger)
  • Linux binaries
    • x86: acsegmentor-0.1-Linux-x86.tar.bz2
    • amd64: acsegmentor-0.1-Linux-amd64.tar.bz2
  • MacOSX binaries
    • Intel and X11: acsegmentor-0.1-MacOSX11-intel.tar.bz2
  • Windows binaries, run it with command.com or Msys to provide command line arguments: acsegmentor-0.1.exe

Contact: burrus@ensta.fr

Contact: burrus@ensta.fr

Contact: burrus@ensta.fr

• Introduction
  • Acsegmentor aims at segmenting an image into homogenenous regions.
  • Use `acsegmentor —help’ to find the (numerous) available command line options
• Segmentation examples

  $ acsegmentor --image lena.pgm --output output.pgm
  => segment the image lena, resulting segmentation is in `output.pgm'.
  => use default parameters: a contrario segmentation with MergeBest 
     exploration heuristic and watershed initialization with sigma=0.8

  $ acsegmentor --image lena.pgm --output output.pgm --initializer egbis
  => idem as above with an egbis initialization with default parameters
     (k=50, sigma=0.8).

  $ acsegmentor --image bigimage.pgm --output output.pgm
  => Exception raised because no thresholds can be found.
  => Thresholds were pre-computed for some standard image sizes 
     256x256, 384x256 and 512x512. For other sizes, you have to
     learn new thresholds (see LEARNING EXAMPLES).

  $ acsegmentor --image bigimage.pgm --output output.pgm --sigma 1.1
  => Exception raised because no thresholds can be found.
  => Specific thresholds must be computed for each parameter. 
     Pre-computed thresholds are given only for default parameters.
  => You have to run a learning process.

  $ acsegmentor --image bigimage.pgm --output output.pgm \
                --threshold-database thresholds.xml
  => Read additional thresholds from `thresholds.xml'. These
     thresholds can be generated automatically by the learning phase.

• Learning examples

  $ acsegmentor --threshold-database-output thresholds.xml \
                --width 1024 --height 1024 --initializer egbis \
                --nb-iterations 1000
  => Learn thresholds for 1024x1024 images for an egbis initializer.
  => 1000 images will be analyzed to deduce statistical thresholds.
  => Resulting thresholds will be put into `thresholds.xml'. This file
     can then be used to segment new 1024x1024 images.

Examples

Two algorithms are implemented to provide to initial partitions: classical watershed (Vincent & Soille) and the algorithm of P. Felzenszwalb and D. Huttenlocher published in the IJCV paper “Efficient Graph-Based Image Segmentation” (authors’ webpage), called EGBIS here.

More details can be found in this draft PDF submitted to Pattern Recognition.

Two algorithms are implemented to provide to initial partitions: classical watershed (Vincent & Soille) and the algorithm of P. Felzenszwalb and D. Huttenlocher published in the IJCV paper “Efficient Graph-Based Image Segmentation” (authors’ webpage), called EGBIS here.

Two algorithms are implemented to provide to initial partitions: classical watershed (Vincent & Soille) and the algorithm of P. Felzenszwalb and D. Huttenlocher published in the IJCV paper “Efficient Graph-Based Image Segmentation” (Authors’ webpage).

Acsegmentor comes with an example of distance function which analyzes frontier contrast, gray level distributions and deviations differences between couples of regions. It also comes with a simple exploration heuristic, which, starting with a initial partition, iteratively merges the least different regions until only meaningful couples of regions remain.

Two algorithms are implemented to provide to initial partitions: classical watershed (Vincent & Soille) and the algorithm of P. Felzenszwalb and D. Huttenlocher published in the IJCV paper “Efficient Graph-Based Image Segmentation” ([[http://people.cs.uchicago.edu/~pff/segment/|Web page of the authors]).

Acsegmentor aims at segmenting gray-level images into homogeneous regions. It’s a generic algorithm: it requires a distance function between regions, an exploration heuristic and automatically find thresholds ensuring that regions in the final partitions will be statistically meaningfully different. This step is done using a contrario reasoning: two regions are meaningfully different is the probability that the exploration heuristic find such different regions in pure noise is very low. This ensures that in the final partition, statistically, regions difference are not due to chance but are a result of some physical phenomena, such as different objects in the scene.

    http://burrus.name/pub/files/research/acsegmentor/house-watershed-filtered.png Filtered (xx regions)

    http://burrus.name/pub/files/research/acsegmentor/house-watershed.png Watershed (xx regions)
    http://burrus.name/pub/files/research/acsegmentor/house-watershed-filtered.png” Filtered (xx regions)

    http://burrus.name/pub/files/research/acsegmentor/house-egbis-filtered.png Filtered (xx regions)

Acsegmentor uses C Make (http://www.cmake.org) to generate Makefiles, K Develop, X Code or Visual Studio projects. Once cmake is installed, you just need to run it on the root directory of the source tree. It is available in most Linux distributions.

Acsegmentor also depends upon VXL (http://vxl.sourceforge.net), Boost (http://boost.org) and Juce (http://www.rawmaterialsoftware.com/juce/). These libraries are embedded in the source tree to avoid extra installations and to make sure that you have a compatible version.

• CMake \\

• Dependencies \\

• Example on Unix \\

• CMake

• C Make

• Dependencies

• Example on Unix

Requirements

Acsegmentor uses C Make (http://www.cmake.org) to generate Makefiles, K Develop, X Code or Visual Studio projects.

Once cmake is installed, you just need to run it on the root directory of the source tree. It is available in most Linux distributions.

Acsegmentor also depends upon VXL (http://vxl.sourceforge.net), Boost (http://boost.org) and Juce (http://www.rawmaterialsoftware.com/juce/).

These libraries are embedded in the source tree to avoid extra installations and to make sure that you have a compatible version.

Example on Unix

$ mkdir build
$ cd build
$ ccmake /path/to/acsegmentor-0.1-Source
=> Console based-interface
=> Type 'c' (configure, maybe twice) 
   and then 'g' (generate) to create Makefiles.
=> You might want to change build settings. For example, to enable
   both optimizations and debugging symbols, 
   enter `RelWithDebInfo' in CMAKE_BUILD_TYPE. For the fastest binaries,
   enter `Release'.

$ make

$ src/acsegmentor
=> should be ok

Acsegmentor aims at segmenting gray-level images into homogeneous regions. It’s a generic algorithm: it requires a distance function between regions, an exploration heuristic and automatically find thresholds ensuring that regions in the final partitions will be statistically meaningfully different. This step is done using a contrario reasoning: two regions are meaningfully different is the probability that the exploration heuristic find such different regions in pure noise is very low. This ensures that in the final partition, statistically, regions difference are not due to chance but are a result of some physical phenomena, such as different objects in the scene.

Acsegmentor comes with an example of distance function which analyze gray level distributions difference, deviations and frontier contrast between couples of regions. It also comes with a simple exploration heuristic, which, starting with a initial partition, iteratively merges the least different regions until only meaningful couples of regions remain.

More details can be found in this draft submitted to Pattern Recognition.

Introduction

Project description

Examples

Download

Installation

Basic usage