[SOLVED] nodes for functions

Hi Christian,

Would it be possible to list nodes(or combos) that may correspond to the following functions?

  • merge images with different focal length
  • metrics for blurriness (contrast, correlation, Tamura kurtosisof directionality ?)
  • de-blurring (sharpen in Fiji?, CLAHE, erosion?)
  • grayscale and color calibration

 

Hi,

see below for the list of nodes. Note, that you can run most of the FIJI Plugins using the ImageJ-Macro Node from the ImageJ Integration. If something doesn't work with the IJ Macro Node or you want to have a certain plugin as a dedicated KNIME node, this is actually a very easy task. Especially if the plugin is an ImageJ2 Plugin. We can automatically create KNIME Nodes from headless ImageJ2 Plugins.

Greetings,

Christian

  • merge images with different focal length

I don't know exactly what you want to do. But we have the following nodes: Merger, Groupby, Image Calculator, etc.

  • metrics for blurriness (contrast, correlation, Tamura kurtosisof directionality ?)

Image Segment Features / Segment Features: Tamura, Haralick, FirstOrderStatistics. In the next release we will have many more features.

  • de-blurring (sharpen in Fiji?, CLAHE, erosion?)

ImageJ Macro Node (Sharpen, CLAHE, Rolling Ball,...), CLAHE, Morphological Operations (Erode/Dilate/Open/Close with or without Structuring Element Creator), H-Dome, Difference of Gaussian, Kernel Creator, Convolver, Image Calculator etc.

  • grayscale and color calibration

What do you mean? Node like: Image Normalizer (Histogram Equalization, Saturation etc), Converter (convert between types and so on), Projector (e.g. Project Color Channels), Image Cropper, etc.

Something missing? We have a lot of nodes and especially the combination of these nodes makes the plugin so powerful. In our tutorials on: http://tech.knime.org/forum/knime-image-processing/ you find some very good examples.

Maybe one more comment: We (KNIME + ImageJ2/FIJI) are currently working on a framework called ImageJ-Ops, where we reimplement all our algorithms and merge them with the existing ones of ImageJ2 such that we have the same code-base in the near future. So everything you get in ImageJ2/FIJI you will automatically get in KNIME as a node (there are exceptions, but for most of the algorithms this statement holds).

Christian

 

Hi,

I would like to measure sharpness (kurtosis, psnr, mtf … or better algorithms that are in the features node??? ) of the follow recorded images:

https://www.dropbox.com/s/bvz4wbkhwinl0k7/IP-Create%20Image%20Subset-02.czi?dl=0

They are microbeads and in reality perfect dots. I would need the measurements for every bead.

What would be a smart way to do that in knime? (Or is there already a good example workflow for that ?)

Kind regards,

Flo

Hi Flo,

1. Step: Spot Detection / Segmentation of Spots (see Spot-Detection Example)

2. Step: Image Segment Features applied on Image + Labeling (see High-Content Screening Workflow as an example -> Feature Calculation).

Does this help?

Christian

ah that Spot Detection example workflow helped a lot. thank you :)

 

One question regarding to mathematics:

Kurtosis and Skewness is defined in 1D right? But the segments are in 2D so what is the node actually calculating?     (Probably a dumb question :D )

The first order statistics are calculated based in the intensity values.

Cheers,

Christian