search

Topp-Roots-Lab / 3d-root-crown-analysis-pipeline

A pipeline to process 3-D volumes (.raw) of maize root crowns. Also known as XRCAP.
3 stars 1 forks source link

New thresholding method for segmentation #11

Closed tparkerd closed 4 years ago

tparkerd commented 4 years ago

Because the bottom of x-ray volumes that were scanned in air (no soil), the histogram is not very wide and the triangle method may select the wrong threshold value, as it depends on the 'shape' of the histogram.

I'm going to explore what can be done to prevent this method from selecting a bad threshold value. Currently, it appears that the issue arises if the threshold value if below the mode grayscale intensity value for the slice.

Two conditions could be that the threshold value is greater than the mode value and that the image is not some percentage of white pixels (maybe 70%). It might be worth collecting the percentage for a large portion of "good" segmented slices.

tparkerd commented 4 years ago

If we work under the assumption that the object of interest is in the center of the volume, then maybe it might be worth applying a mask of a circle gradient to each slice to augment the actual values that are root system instead of air/soil. I think this method may help with root crowns that are solely in air. It would however have the same effect on any embedded soil.

https://stackoverflow.com/questions/46466563/fill-circle-with-gradient/46467000

tparkerd commented 4 years ago

Resolved by #18