Our FR10 inspection examples show a bright line wandering across the same bezel as its angle changes, and somebody has labelled every image scratched. Still have the bezel, no agreed physical verdict. Can we sort that on the bench before buying more lighting?
Who inspected the bezel before it became a reject reference? Find that person or ask quality to examine it now. I would leave the images unclassified until the physical part has a clear status.
Then the image label is where to start. Keep the same part available for a comparison under controlled viewing and presentation. You can investigate the moving reflection while quality decides whether the surface is acceptable.
Quality has it now, along with two known good bezels. They've asked us to stop filling the reject folder until the samples are agreed. Fair request, slightly embarrassing folder.
Did they find a physical mark at the bright line's original position? I'm interested in whether you had reflection alone or reflection over a real defect.