Half-discs samples are supposed to have their pith centred on a particular spike. It would be helpful if this were checked at scan collection time and the operator advised.
Possibly the pith is most easily located in the XRA data.
To do this it might help to label the pith spike clearly in the XRA image (e.g. by drilling a small hole through the cassette rail).
It may not be necessary to actually find the pith, but rather warn operator if the pith dosn't appear to be in the vicinity of the pith spike.
Maybe:
extract a sub image centered on the pith spike
use hough transform to locate centre of any circular features, or
edge enhance then compute intersection of edge fragment normals
compare center position with nominal.
compute some sort of centre-finding credibility measure
if either credibility too low or position unacceptable warn operator
This algorithm might also be useful in refining the pith position.
DB0088 [Created: 2017-01-20]
Half-discs samples are supposed to have their pith centred on a particular spike. It would be helpful if this were checked at scan collection time and the operator advised.