KNIP - KNIME Image Processing

Fri, 12/03/2010 - 13:09 — knime_admin

These image processing nodes add new image types to KNIME and the corresponding nodes to read more than 100 different kinds of images (thanks to the Bio-Formats API), to apply well known methods for the preprocessing and to perform image segmentation. It replaces the outdated Image Processing plugin that was previously hosted on KNIME Labs. Most of the included nodes operate on multi-dimensional image data (e.g. videos, 3D images, multi-channel images or even a combination of them), which is made possible by the internally used ImgLib-API.

In addition several nodes are included to calculate image features (e.g. zernike-, texture- or histogram features) for segmented images (e.g. a single cell). These feature vectors can then be used to apply machine learning methods in order to train and apply a classifier.

:: Overview ::

  • KNIPLib: KNIME independent ImgLib2 based image processing framework
    • Includes about 120 operations containing functionality for preprocessing, segmentation, classification (features), tracking, annotating, viewing etc. n-dimensional images
    • Bases on: ImgLib2, BioFormats for reading images
  • KNIPPlugin: KNIME plugin to use functionality of KNIPLib within KNIME
    • >70 nodes for segmentation, feature extraction, preprocessing, filtering, viewing, labeling ... n-dimensional images
    • >30 nodes in different not yet public side projects (tracking, Pixelbase classification, imageJ integration, VTK based 3D Viewer)

:: Nightly builds of current plugins ::

For installation instructions please click here.

:: Example Workflows ::

To run the workflows please install the Math-Formula plugin to also see the analysis. This is not needed, but a error will be thrown if it is not installed. the workflows run anyway!

:: News ::

2011-01-30

Report from IJ2 Hackathon:

  • Martin worked hard on the ImageJ2 integration. Not finished yet but it should be ready with the first offical release of IJ2
  • Christian worked on the integration of KNIPLib operations with ImgLib2 operations (and of course all the algorithms)! We will share more code in the future and try to integrate even more with ImgLib2. Work in progress.
  • We will soon have a prototype of a new OMERO Reader/Writer. Stay tuned. 
  • It was really cool to meet all the people from Madison, Dresden, Dundee, etc.. and we really want to collaborate even more in the future! Thanks again to Pavel!

KNIP - 3D Viewer Beta

Major changes in KNIPLib + Plugin

  • As we want to have a V1.0 at some time, we need to make some changes now concerning the organization of nodes in the plugins and the operation concept (see IJ2 Hackathon report).
  • These changes will pretty likely have effects on your existing workflows and nodes which you then have to adapt. We will make a bigger announcement in the next weeks about this issue!
     
2011-09-12

Some new nodes / new RegionGrowingFramework:

  • We added some new nodes: Grow Labeling, Shrink Labeling, Extract Outline from Labeling
  • We revised the "Voronoilike Segmentation" Node
  • We developed a RegionGrowingFramework for KNIPLIB which makes it possible to quickly implement arbitrary regiongrowing algorithms.
  • We are working hard on the documentation, stay tuned.
  • We will go to Dresden tomorrow, let's see what we can do there ;)

Cheers,

M&C
2011-28-11

Several changes / bugfixes:

  • In the last weeks we were busy bugfixing a lot of stuff. For example critical annotator and image reader bugs were fixed as well as some other smaller bugfixes in several nodes.
  • Some nodes (e.g. Table to Image, Table to Image, Converter,...) were enhanced with new functionality. The converter can now directly normalize an image. Table to Image and Labeling to Image can now create images in with arbitrary pixel formats.
  • We are working on a documentation which allows and quickstart into using and developing KNIP nodes. This documentation will contain example workflows as well as example nodes with the source code to make it easier for new users to get started with KNIP. There will also be a section about Images, Labelings, FactoryTypes etc.
  • Handling large amounts of images (e.g. from High-Throughput-Screenings) can no be done by using the VM-Argument:  -Dorg.knime.container.cellsinmemory=X where X corresponds to the number of images which you can keep in your main memory. If you are not sure what to chose her just set X = 1. We are working on an easier and especially faster solution which hopefully will be available in the first month's of 2012
  • We are going to the IJ2 Hackathon in Dresden next week. Stay tuned about the results of the integration of IJ2 into KNIME!
2011-06-10

We have changed our name to KNIP (Konstanz Image Processing) and also our SVN structure. You can get the source from:

https://svn.uni-konstanz.de/incide/knip
user: testusr, pw: testitnow

There you need to check out trunk/library/org.kniplib and trunk/plugins/org.knime.knip.base.

If you have any bugs to report, suggestions about the library and/or the framework or question about how to use/install the KNIME plugin, please use our mailing list (knip-devel@mailman.uni-konstanz.de). Please register under: https://mailman.uni-konstanz.de/mailman/listinfo/knip-devel or visit us on irc.freenode.net in channel #knip
2011-29-09 Stable release postponed to the time after des KOS-Days. Until then please use the nightly built.
2011-09-01  Beta version of  revised KNIME Image Processing Plugin available (available http://tech.knime.org/update/community-contributions/nightly for the nigtly build, see installation instructions). We would be pleased about bug reports and suggestions for improvements.

 

Contact:

Martin Horn Nycomed Chair for Bioinformatics and Information Mining and Research School Chemical Biology, University of Konstanz
Christian Dietz Center for Applied Photonics (CAP), University of Konstanz

 

Screenshots: