How should I compare cover images when the seating changes?

CarlaAbbott0053 · 2025年6月27日 17:48 UTC

已关闭

主帖中文翻译暂不可用,以下显示英文原文;回复保留原文。 查看英文原文

CA
CarlaAbbott0053
I tried a simple positioning reference on an earlier bench comparison and it helped us return to the same view. Our new FR5 cover-inspection study has a more awkward problem: the polished cover can sit on its rim or on the intended supports, and the image changes enough to produce a bright line. Before we buy different lighting, I want to compare repeat captures with the cover properly seated. What should I include so we do not merely choose the arrangement that hides a real mark?

11 条回复

SO
SofiaBaker0456

Do you have a cover with a confirmed reject mark for the comparison?

6分
CA
CarlaAbbott0053

Yes, one rejected for a scored patch and two accepted covers. Quality has identified the actual areas. The disputed bright line is on one of the accepted covers.

11分
SO
SofiaBaker0456

I'd include all three, keeping each linked to its physical sample.

9分
DA
DavidBaker0519

Fix the seating problem first. Otherwise your lighting comparison changes two things at once.

-9分
CA
CarlaAbbott0053

That is our plan. I still want images of the wrong seating as a comparison, but not mixed into the repeatability results for the intended position.

20分
SA
SamBennett0768

I'd keep those wrong-seating images for whoever writes the operator messages too, because 'image quality poor' won't tell them they've perched a cover on its rim.

17分
SO
SofiaBaker0456

Would the system actually detect that seating error, Sam?

2分
CA
CarlaAbbott0053

Not currently. Sam's suggestion is useful as a training example, but we cannot promise a message from the camera yet. Fixture designer is reviewing how to prevent the false seating.

8分
IM
ImranAdams0131

Our training examples once included a failure the real screen could not identify. Operators waited for a warning that never came. Label that distinction clearly

4分
CA
CarlaAbbott0053

Designer has proposed a change to the rim clearance. We have not tried it yet. Thanks for catching the message assumption; the comparison sheet now separates physical seating, observed image and any actual detection.

3分
SA
SamBennett0768

Fair correction, I jumped from an image that shows the problem to a camera that knows what it means.

20分

讨论已关闭

此讨论已连续六个月没有活动,现已关闭新回复。 最后活动: .