How to create LabelToRGBFunctor

Beginner Questions
1

for the C++ implementation of LabelMapToRGBImageFilter
It says to use a functor to the colormap,

im simpleITK, they convert a colormap into a functor:

However, in the python version how to I create a type of itk::Functor::LabelToRGBFunctor?

Thanks

2

It does not appear that the LabelToRGBFunctor class is wrapped yet:

This can be easily done by adding a new wrapping for the class. Please see the ITK Software Guide for more information on how to do this.

3

hmm ya it seems so. how do you set the color map then?

4

It looks like many of the filters in the ImageFusion module provide the following interface:

 /** Add color to the LUT container */
  void AddColor(ComponentType r, ComponentType g, ComponentType b);

  /** Empty the color LUT container */
  void ResetColors();

  /** Get number of colors in the LUT container */
  unsigned int GetNumberOfColors() const;
5

hmm,
I need to set the colours for itk.LabelMapOverlayImageFilter though,
and it only has SetFunctor, any idea how I can do this?

I can call .GetFunctor() but it returns a
<Swig Object of type ‘itk::Functor::LabelToRGBFunctor< unsigned long,itk::RGBPixel< unsigned char > > *’ at 0x12e007c30>

and I have no idea how to edit it in python

also, how do I access the image fusion module?

6

Hi @benjaminhon,

Here is an example:

 #!/usr/bin/env python

import sys
import itk

if len(sys.argv) != 4:
    print("Usage: " + sys.argv[0] +
          " <InputFileName> <LabelMap> <OutputFileName>")
    sys.exit(1)

inputFileName = sys.argv[1]
labelFileName = sys.argv[2]
outputFileName = sys.argv[3]

PixelType = itk.ctype('unsigned char')
Dimension = 2

ImageType = itk.Image[PixelType, Dimension]


reader = itk.ImageFileReader[ImageType].New()
reader.SetFileName(inputFileName)

labelReader = itk.ImageFileReader[ImageType].New()
labelReader.SetFileName(labelFileName)


LabelType = itk.ctype('unsigned long')
LabelObjectType = itk.StatisticsLabelObject[LabelType, Dimension]
LabelMapType = itk.LabelMap[LabelObjectType]

converter = itk.LabelImageToLabelMapFilter[ImageType, LabelMapType].New()
converter.SetInput(labelReader)

RGBImageType = itk.Image[itk.RGBPixel[PixelType], Dimension]
overlayFilter = itk.LabelMapOverlayImageFilter[LabelMapType, ImageType, RGBImageType].New()
overlayFilter.SetInput(converter.GetOutput())
overlayFilter.SetFeatureImage(reader.GetOutput())
overlayFilter.SetOpacity(0.5)


itk.imwrite(overlayFilter.GetOutput(), outputFileName)
7

Hi Matt,
But how would you set the colormap for the labelmapfilter?

I ended up using LabelToRGBImageFilter instead of LabelMapToRGBImageFilter

However, I realise that I can only ResetColors() once, which means I can only set the color at the first time. After that I cannot dynamically change the colours, calling ResetColors() again and AddColor() does not seem to change the image.

I did further tests,

it seems that for OverlayLabelFilter, changing colours dynamically(ResetColors(), AddColor) works, but not for LabelToRGBImageFilter. Maybe this is a bug?

8

What happens if you manually call Update() after a call to ResetColors() or AddColor()?

9

itk.LabelMapOverlayImageFilter currently has a fixed progression of colors.

For LabelToRGBImageFilter or LabelMapToRGBImageFilter, is there a simple, self contained example we could run to reproduce the behavior observed?

10

Hi Matt, here are my functions, ill try to make a self contained example. Actually both overlay and labelmap does not seem to allow Reseting the colors and adding new colors.

Hi Dzenan, yep I’ve called Update()

def LabelToRGBImageFilter(input, colors=[]):
    inputType = type(input)
    labelToRGBImageFilter = itk.LabelToRGBImageFilter[inputType, itk.Image[itk.RGBPixel[itk.UC], 3]].New(input)
    if len(colors) > 0:
        labelToRGBImageFilter.ResetColors()
        for color in colors:
            labelToRGBImageFilter.AddColor(*color)
    labelToRGBImageFilter.Update()
    return labelToRGBImageFilter

def OverlayLabelFilter(featureImage, labelImage, opacity=0.5, colors=[]):
    labelType   = type(labelImage)
    featureType = type(featureImage)
    labelOverlayImageFilter = itk.LabelOverlayImageFilter[labelType, featureType,itk.Image[itk.RGBPixel[itk.UC], 3]].New()
    labelOverlayImageFilter.SetInput(featureImage)
    labelOverlayImageFilter.SetLabelImage(labelImage)
    labelOverlayImageFilter.SetOpacity(opacity);
    if len(colors) > 0:
        labelOverlayImageFilter.ResetColors()
        for color in colors:
            labelOverlayImageFilter.AddColor(*color)
    labelOverlayImageFilter.Update()
    return labelOverlayImageFilter

after that, when trying to reset the colors

labelOverlayImageFilter.ResetColors()
for color in [(0,0,0), (255,0,0),(0,255,0),(0,0,255)]:
            labelOverlayImageFilter.AddColor(*color)
1 Like