Dimensional inspection: diagnose bias and drift

Inspection task

First determine whether the error is fixed bias, position-dependent scale or time-dependent drift. Replaying the same image isolates processing changes from acquisition changes. Use an independently measured reference; a repeated numerical result is not proof of truth. Keep calibration validity failures separate from ordinary out-of-tolerance parts.

Prerequisites and sample controls

Define the physical dimension, datum, permitted plane and unit before processing an edge. Provide a traceable reference artifact and independent dimensional answers. A visible contour or a pixel length is not a calibrated measurement. The current prepared native measurement route binds calibration and approved materials; missing approval, mismatched identity or an invalid plane must prevent a valid physical result. Hardware and production acceptance remain installation-specific.

Choose the camera route

For planar edge measurements, compare a low-distortion lens with a telecentric route where height variation changes perspective. Select an area-scan sensor that resolves the edge across the full required field. Height and profile require a verified 3D route rather than a scalar pixel-to-millimeter conversion. Changing lens, focus, camera position or measurement plane requires a calibration validity review.

Run the controlled procedure

Freeze a copy of the failing workflow and retain the original failed image before testing a correction. Reproduce the symptom on saved data first, then determine whether it follows the part, illumination, acquisition or software state. A changed final answer is insufficient: the relevant intermediate evidence must improve for the expected physical reason. Do not erase the initial failure after a successful rerun.

1. Replay the failed image with its original calibration identity.
2. Inspect which edges and points formed the dimension.
3. Move the reference across permitted positions in new captures.
4. Check focus, plane and mounting before changing tolerances.

Record the evidence

Record pixel geometry, artifact answer, reported dimension and difference. State whether each observation uses the same saved image or a newly captured frame. Retain both before/after results.

Build a residual ledger around an independently measured artifact. For every observation, store the original image points, the reported dimension, the independent reference answer and the difference in the declared unit. Group the observations by field position, acquisition time and physical presentation rather than combining all errors into one average. Replay one original without changing materials to check whether the processing result itself is stable. Then acquire new images of the same artifact at different permitted positions. A position-dependent difference suggests a spatial applicability issue; a shift after a lens or fixture change suggests that the previous calibration context no longer describes the installation. Keep these as diagnostic hypotheses until the relevant controlled comparison supports them.

Diagnose failed cases

A wrong unit or calibration binding can produce a stable large bias. Height and lens changes can create location-dependent error. Adjusting the tolerance only hides disagreement; restore valid geometry and calibration instead.

Validate independently

Alter one suspected cause per trial and replay all previously failing controls. Keep image IDs, parameter differences, ROI and runtime identity in the log. If the result changes without a corresponding change in the expected intermediate observation, investigate timing or provenance. Inconclusive frames should remain explicit review cases instead of being removed to make a comparison look better.

Verify correction on another independently measured artifact and the original failing positions. If approved materials are unavailable or mismatched, refuse a valid measurement rather than returning a plausible fallback.

Separate loss of validity from a valid but incorrect dimension. If the calibration identity or approved plane does not match, stop the metric route and record that refusal; do not calculate a residual from an invented fallback number. If geometry is missing, retain the review state and inspect raw edge evidence. For a valid measurement with bias, test edge selection, optical presentation and reference integrity one at a time. Do not recalibrate against the same artifact and then use that artifact alone to prove the correction. Reserve another independently measured artifact and previously unused positions for the confirmation step. Preserve the old calibration, old failures and new candidate context with their identities. The final diagnosis should explain which physical or computational boundary changed, why the residual improved and which conditions remain outside validity. A tolerance expansion is a specification decision requiring approval, not a substitute for that diagnosis.

Continue the engineering handoff

Send the original failure, corrected result, parameter difference and remaining boundary cases to Deyi Vision for review. Verify corrections on the live installation only after saved-image isolation succeeds. A correction demonstrated on one file is not evidence of physical-camera reliability, robot calibration or production acceptance.

Save the original image and the resulting record together.