Pipe internal surface inspection: controlling glare

A reflective pipe bore can look bright and detailed while concealing the defects you need to reject. The difficulty is not simply excessive light: illumination angle, restricted access and changing image scale can all hide a scratch or pit. For an automated inspection trial, establish defect visibility and surface coverage before choosing hardware.

Start with representative parts, including acceptable finish variations and confirmed rejects. Compare lighting geometries at several depths, then measure the smallest visible defect and the surface actually inspected. This separates an attractive demonstration image from a repeatable inspection method.

Define what the inspection must reject

Write the acceptance criteria before adjusting exposure. “Inspect the internal surface” is not specific enough to establish optical resolution, lighting or coverage.

For each defect class, record:

• Minimum rejectable width and length, with depth where relevant.
• Axial location and circumferential orientation.
• Whether the defect changes reflectivity, surface slope or both.
• Acceptable machining marks, weld features and finish variation.
• Whether the bore must be inspected completely, including entrance and exit transitions.

For example, detecting a dark deposit is a different task from finding a shallow scratch in polished metal. A scratch may appear under directional illumination but disappear when its orientation changes. A pit may produce a bright edge beside a dark region rather than a uniformly dark spot.

Build the trial around identified defects, not unidentified bright pixels. Use measured or otherwise characterized samples, and keep separate examples for final validation. Artificial marks can help compare lighting, but they do not automatically represent production defects.

Also establish the required cycle time and whether the pipe can stop, rotate or accept an inserted optical head. These constraints determine which viewing arrangements deserve testing.

Prove optical access before selecting resolution

An end-on view does not automatically inspect the entire internal wall. A long, narrow bore produces increasingly grazing views at greater depths, with strong foreshortening and difficult illumination paths.

Consider an illustrative bore with a 20 mm internal diameter and a 300 mm length. Its length-to-diameter ratio is:

For an idealized viewing point centered at the entrance, a ray reaching the far wall makes approximately this angle to the bore axis:

This estimate does not model a real lens. It illustrates why distant wall features are viewed almost along the surface, rather than face-on.

Compare three access arrangements:

A circumferential optical arrangement may reduce rotational scanning, but it does not establish full-length coverage. You may still need controlled axial translation.

The inspection-optics guide covering inner walls and bores explains the differences between optical arrangements. Use the inspection-lens category and lens finder only after defining the surface each image must cover.

Compare scattered light with controlled specular reflection

Do not make “remove every reflection” the objective. Specular reflection can conceal a defect, but a controlled reflection can also reveal a change in local surface slope.

Low-angle illumination can make scratches and roughness scatter light toward the camera while the smoother background remains dark. Whether this happens depends on the local wall normal and the complete illumination-to-camera geometry. In a cylindrical bore, those relationships change around the circumference.

The following diagram illustrates the local bright-field and dark-field principle, not a complete bore-inspection layout.

Test the approaches separately before combining them:

Keep exposure, gain and processing fixed when comparing illumination directions at a given setup. Record whether defects become brighter or darker than their surroundings; a single threshold polarity may not suit every condition.

Cross polarization is worth testing, not assuming. Its effectiveness depends on material, viewing geometry and reflections within the bore. The cross-polarization guide explains the practical limitations.

A dome can help with accessible reflective surfaces, but restricted access changes what illumination reaches a deep bore. Use the diffuse-dome lighting guide to understand that distinction rather than assuming external diffuse lighting reaches every internal location.

Calculate local sampling, not just camera megapixels

Choose a provisional sampling target from the smallest defect, then verify it experimentally. Four pixels across a defect is a useful starting point for a visibility trial, not a guaranteed detection specification.

Suppose the minimum target width is 0.10 mm:

This suggests roughly 2,600 usable circumferential samples as an initial trial target, allowing a small sampling margin. It does not establish the required sensor dimensions or guarantee optical contrast.

The resolution calculator can estimate the linear sampling requirement when you enter the developed circumference as the inspection width. Treat that as a planning approximation, not an annular-optics model.

An annular image does not use its pixels like a rectangular image. Circumferential sampling changes with image radius, while axial scale depends on the optical mapping. Unwrapping can create more output pixels without adding resolved detail.

Measure local scale using a suitable reference at several depths and angular positions. Check the least favorable location, including any regions obscured by supports or affected by reduced sharpness. Evaluate defect visibility in the original image as well as the unwrapped image.

If contrast remains poor despite adequate sampling, investigate stray-light causes of low image contrast before increasing pixel count.

Work through axial coverage and motion

After proving a useful circumferential view, measure the axial wall band that meets your detection requirement. Exclude blurred edges and regions where glare makes the target defect unreliable.

Suppose that validated band is 12 mm long. With 25% overlap, the axial increment becomes:

This calculation assumes that the first valid band starts at the beginning of the required inspection region. Entrance geometry, positioning uncertainty or degraded edge quality can require additional travel and captures.

Overlap provides coverage margin and supports image registration. It cannot compensate for a blind angular sector repeated in every frame. Validate the entire circumference at every axial region.

For continuous translation, budget motion blur using the measured local scale. At 50 mm/s, allowing 0.5 pixel of blur with 0.025 mm/pixel sampling gives:

This assumes axial motion maps directly to the measured axial image scale. Rotation needs its own calculation using tangential surface speed.

A stop-and-image trial simplifies the initial optical comparison. Later, validate exposure, vibration settling and motion timing under the intended production cycle. If pulsed illumination is needed, follow the strobe timing and duty-cycle guide; supply voltage alone does not establish safe pulse operation.

Run a controlled imaging trial

Use a small test matrix that separates optical limitations from processing limitations.

1. Establish a stationary baseline. Hold the pipe and imaging head securely. Record axial position, angular orientation, focus, exposure and gain.
2. Capture acceptable surface variation. Include representative finishes and ordinary marks that must pass. Otherwise, the trial can favor lighting that makes everything look defective.
3. Test known rejects across the bore. Examine entrance, middle and far-end regions, with additional positions wherever the image changes significantly.
4. Change one lighting variable at a time. Compare direction, elevation and intensity separately. Keep unsaturated originals for every condition.
5. Test defect orientation. Present longitudinal, circumferential and oblique scratches where those classes matter. A successful view of one orientation is not sufficient evidence.
6. Repeat after unloading and reloading. Check whether ordinary positioning variation changes glare, scale or coverage enough to affect detection.
7. Add production motion last. Once stationary visibility is established, repeat under the intended translation, rotation and capture sequence.

Save raw images and configuration records before applying contrast enhancement. Inspect saturated-pixel regions directly: clipping can hide both bright defects and the reflection changes that indicate them.

If multiple exposures are necessary, consider motion and registration before combining them. The HDR capture and fusion guide explains why increased dynamic range does not repair missing optical access or a defect that produces no contrast.

Shortlist parts without assuming a complete solution

Catalog parts are candidates for a trial, not evidence that a complete assembly fits inside the bore.

Review LS-PCL-0042 and LS-PCL-0066 as inspection-optics candidates. Confirm the intended viewing arrangement, mechanical dimensions, usable wall coverage and camera compatibility before selecting either.

Compare LS-SPC-0142 and LS-SPC-0177 when investigating multi-angle or segmented illumination concepts. Neither listing establishes that the light can illuminate the required deep-bore surface from your chosen mounting position.

Use the special-lights category and lighting finder to explore arrangements after documenting which angles reveal the defects. Verify channel control, trigger compatibility and optical delivery separately; a segmented construction does not by itself establish the required sequencing behavior.

The MV-CU020-90GC is a global-shutter camera candidate. That shutter type can be useful for motion capture, but local sampling, optical compatibility and sensitivity still require verification.

Compare area scan cameras through the camera finder once the trial establishes the required image geometry and exposure. Do not treat these candidates as a confirmed matched assembly. The lens configuration step is a starting point for organizing the optical requirements.

Know when the usual approach is wrong

More light is not the answer when the useful reflection is already saturated. Reduce illumination or exposure, or change the reflection geometry. Likewise, increased camera resolution cannot reveal a surface hidden from the optical path.

A single lighting direction is unsuitable when rejectable scratches disappear at certain orientations. Test complementary directions and evaluate each capture separately before deciding how to combine the evidence.

A single end-on capture is unsuitable when far-end sampling, foreshortening or reflection patterns prevent reliable detection. Consider an inserted viewing arrangement, controlled scanning or access from both ends. Opposite-end viewing still needs a demonstrated overlap region.

If the defect requirement is fundamentally a depth measurement, intensity images alone may be insufficient. Reflection contrast depends on surface finish and angle, so brightness is not a direct depth measurement.

Finally, do not classify an inaccessible region as inspected. If no practical optical arrangement reveals the required defect there, revise the inspection architecture or the coverage requirement explicitly.

Check these points before ordering

• Defect definition: minimum dimensions, orientations and acceptance limits are documented.
• Mechanical access: the complete optical head, mount and service connections have verified clearance throughout travel.
• Coverage map: every required axial and angular region has demonstrated defect visibility.
• Sampling: local scale is measured at the least favorable image locations, not inferred from total megapixels.
• Lighting evidence: acceptable variations and known defects remain distinguishable without saturated regions hiding information.
• Motion budget: exposure, settling, translation, rotation and capture timing fit the required cycle.
• Compatibility: sensor coverage, mount, focus range, illumination control and triggering are confirmed for the proposed assembly.
• Validation plan: independent samples and repeated loading trials will measure missed defects and false rejects.

Bring representative parts, the coverage map and unprocessed trial images to the next review. Use the lighting configuration step to organize a candidate setup, or contact the applications team with the bore dimensions, defect criteria and access constraints before committing to cameras, optics and illumination.