Hello imageJ people,
I may miss an important point here and I couldn't find an answer anywhere.
Problem
Let's say I have a 32-bit grayscale image and I color an even region using the floodFill(x0,y0) command and then get a pixel intensity using getPixel(x1,y1), where both P(x0,y0) and P(x1,y1) are part of the previously mentioned region. With setForegroundColor(R,G,B) I define a fill color, where R,G,B in [0,255].
Code:
setForegroundColor(R,G,B);
floodFill(x0,y0);
getPixel(x1,y1);
how do I know what the getPixel() function will return for any pixel inside this region? I mean is there an easy way to calculate the relation between f(R,G,B,imageType) and the getPixel() return value for it in general?
Maybe something like:
f(R,G,B,imageType) = n
n = return value of getPixel()
R,G,B = the chosen floodFill values
imageType = {8-bit, 16-bit, 32-bit, RGB Color}
I'm well aware that different RGB values would return the same getPixel() value multiple times in grayscale images, therefore it's a unidirectional relation.
You need to check the color of the pixel whether you already set it or not.
I'm not completely sure, but I think the key may lie in the public method getBestIndex(Color c) within ImageProcessor.java:
http://rsb.info.nih.gov/ij/developer/source/ij/process/ImageProcessor.java.html
I believe this is called from setForegroundColor() and it looks to me like the current lookup table influences the result - making it harder to predict what will happen, unless you've set the LUT yourself.
To avoid conversion when writing a plugin to work on a 32-bit images, I would always use
FloatProcessor.setValue(double value);
to set the pixel values prior to drawing / filling.
The setColor(value) macro function looks similar, and I think would be a better choice.
If that isn't what you need, then getValue("color.foreground") may help you, or else I would try using getPixel(x0, x0) to find out what the value was for the pixel I started the flood fill at. However if you allow conversion from RGB to happen at all, then that seems a bit risky if it is possible that different RGB values can be mapped to the same 32-bit value afterwards...
Thanks for your answer peter. I think that was very helpful! I was considering creating a loop in which i set all possible values with floodFill() and then get the results from getPixel() for each: 8-bit, 16-bit and RGB color. but in hindsight 3*256^3 datapoints seem a bit over the top for answering this question :-) Then I completely forgot about LUTs, so there would be another problem.
I will definitely give it a try with the getValue("color.foreground") method.
Oh, I didn't realize it's you... thank you for writing this document: "Analyzing fluorescence microscopy images with ImageJ" I learned almost all I know about imageJ from reading this! great work, I appreciate it very much!
- Badges
- Science topic
- Similar topics
- Chemistry
- Pharmacology
- Medical Pharmacology
- Clinical Pharmacology