Our proposed FR10 inspection has inherited two faceplate image folders. A bright streak changes position between presentations, and the labels call each position a scratch. I cannot find a record of anyone comparing those labels with the retained physical plates. I would like to repair the reference set before asking the programmer to use it.
Have quality identify the actual plates and marks beside those captures before changing any thresholds; first check that the two folders even use the same plate identities.
The folders used the same short labels for different plates. Quality has separated them and checked the physical samples. One labelled scratch is a real mark; another has no corresponding defect at the indicated location.
It is still weak in one view. The corrected identity index is now shared with the programmer, and the imaging engineer has the marked plate for a new comparison. We have fixed the file confusion, not the visibility problem.