I am segmenting multiphase particles (rocks, cement clinker) using SEM BE and X-ray image sets. Each particle may contain up to 10 phases, including an epoxy that surrounds the particles and fills pores, appearing as the darkest phase in the image set. I want to measure the area fraction and surface area for each phase. I generate a TIF image where each mineral phase has a single, fixed color and index value, with the epoxy, filling pores and surrounding the particles as the darkest (black) phase. Measuring surface area is a simple process of listing the histogram contents. I also want to identify and quantify the layer(s) of pixels surrounding the particles, isolating them has proven more difficult.
I initially created a Binary image, or a Mask, using the Process Binary tools however, I find that both of these operations will not include all the phases within the particles. I do not see a way to set a range of intensity values to include all the minerals. I did try to use the Segment Active Mask plugin but it appears to get stuck in the Processing Image, and I have not succeeded in getting any output.
For the remaining phases though, this works very nicely as I run a single erosion operation on the mask and see a single layer of pixels removed, just like the skin off an onion. I then take the difference between the original and eroded mask and compare it to the original thresholded image (AND operation), generating a layer of pixels with the color and index values for each phase. This works great with the exception that the Mask or Binary excludes some phases.
An alternative approach was to threshold the image and then try the Make Binary or Mask however, when I try to run an erosion operation, I get an extremely coarse layer of at least 60 pixels stripped from the perimeter, too coarse to be practical. I do not see any means to control the severity of the erosion process.
I also do not understand why the epoxy background of my images often gets turned to level 255, whereas I want my particle features at 255 in a binary. This seems counterintuitive, as the epoxy binder, nothing, should be zero level.
How can I include all the phases in my mask or binary image, so I can make use of the single pixel layer erosion?
Thank you, Paul
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
