Polarized lighting for screen inspection: setup and tests

Reflective screens can hide a coating boundary or surface transition beneath a much stronger reflection. Increasing brightness may raise both signals without making the boundary easier to inspect. This application is for engineers examining cover glass, coated panels, or assembled screens with line-scan acquisition. You will learn how to separate illumination geometry from polarization effects, compare analyzer settings, and check whether the improvement survives motion, part variation, and repeated setup.

Decide what polarization must reveal

Use polarized illumination when changing the polarization state produces a repeatable difference between the target transition and acceptable background. Do not assume that crossing two polarizers automatically gives the best inspection image.

A linear polarizer in the illumination path sets the incident polarization direction. A second polarizer, called the analyzer, sits in the camera path. Rotating the analyzer changes which polarization component reaches the sensor.

Some reflections are strongly suppressed with crossed orientations. Other reflections remain because surface angle, scattering, coatings, and layered materials change the returned polarization. A useful defect may become clearer, disappear with the glare, or remain indistinguishable from normal material variation.

Define the inspection target before adjusting either polarizer. Is the required feature a coating edge, scratch, delamination boundary, contamination patch, or change in surface slope? These are different optical tasks. Polarization contrast alone does not identify the physical cause or measure a step height.

The lighting-angle selection guide explains the geometry decision that should precede polarization tuning. First find an angle that reveals the feature; then determine whether polarization improves its separation from the background.

Build a controlled line-scan setup

Arrange the camera so its imaging line crosses the full inspection width. Move the screen perpendicular to that line, and illuminate the strip being imaged. Select line scan lights for the required illuminated width and geometry, using the lighting finder to narrow the candidates.

The trial setup needs these functional elements:

• A camera and lens that resolve the minimum feature across the inspection width.
• A line light positioned to provide useful contrast at the camera's imaging line.
• An illumination polarizer covering the beam that reaches the inspected strip.
• A camera analyzer covering the complete imaging aperture without clipping the field.
• A compatible light driver, stable mounting, and a repeatable motion reference.
• Representative acceptable parts and independently confirmed defect samples.

Check the polarizers' wavelength range, clear aperture, optical quality, and allowable thermal exposure with their manufacturer. The catalog entries supplied here do not establish compatibility with a particular polarizer.

If a diffuser is used, place it before the final illumination polarizer unless the complete optical stack has been characterized. Scattering after the polarizer can alter the polarization state. Protective windows and other downstream optics also belong in the test configuration.

Use the lighting accessories category and accessory finder for relevant mounting and connection parts. Confirm fit and electrical compatibility rather than selecting accessories by appearance.

Keep exposure, gain, aperture, focus, and image processing fixed during each comparison. Automatic exposure can conceal the brightness penalty of an analyzer and make two configurations appear more comparable than they are.

Compare analyzer orientations before choosing a method

Begin with the illumination polarizer fixed. Capture the same stationary sample region with the analyzer at parallel, intermediate, and crossed orientations. For an initial sweep, use 0, 30, 60, and 90 degrees relative to the source polarizer, then refine around promising settings.

For ideal linearly polarized light passing through an ideal analyzer, Malus's law gives:

Here, theta is the angle between the light's polarization direction and the analyzer transmission axis. The equation predicts half the parallel intensity at 45 degrees and zero at 90 degrees. Those are ideal transmitted-light values, not predictions for a complete reflective-screen assembly.

A real screen can introduce scattering, polarization rotation, phase changes, and reflections from several interfaces. Crossed transmission will therefore not necessarily approach zero. Measure the actual background and feature signals instead of calculating them from analyzer angle alone.

Start with the simplest method that passes validation. A fixed analyzer is usually easier to integrate than combining several measurements from a moving surface.

Calculate sampling and exposure for a moving screen

Consider an application example with an inspection width of 240 mm, a camera line containing 4096 active pixels, and travel speed of 300 mm/s. These are example requirements, not specifications for a catalog product.

The object-space sampling across the screen is:

This produces approximately equal sampling in both directions at constant speed. It does not establish that the smallest defect is detectable; the lens, contrast, noise, and required number of pixels across the feature still matter. Use the line-scan field-of-view guide to establish the optical coverage first.

At an exposure of 50 microseconds, the screen travels:

That blur estimate is a useful starting point, not a universal acceptance limit. Verify it against the narrowest feature and the inspection algorithm. Polarizers reduce available light, so a stationary image that looks excellent may fail at the exposure required for motion.

The line-light power planning guide addresses that light budget. The line-rate setup guide covers acquisition timing and encoder-based sampling when speed changes.

Alternating analyzer channels creates another sampling constraint. If two orientations alternate at a total rate of 5120 lines/s, each orientation receives only 2560 lines/s. Its travel-direction sampling becomes approximately 0.117 mm at the example speed.

Increasing the total rate to 10240 lines/s restores approximately 0.0586 mm sampling within each channel. However, adjacent measurements still come from different positions. Registration remains necessary; doubling the rate does not create simultaneous observations of the same point.

Analyze contrast without confusing it with brightness

For every candidate setting, measure a representative feature region and an acceptable background region. Avoid saturated pixels and include background variation from coating texture, dust, and ordinary manufacturing variation.

A useful screening metric is contrast-to-noise ratio:

Use the same region-selection method throughout the trial. This metric helps rank setups but does not replace a defect-detection study.

For example, suppose a parallel image has feature and background means of 180 and 160 digital counts. If both standard deviations are 4 counts, CNR is 20 / sqrt(16 + 16), or about 3.5.

A crossed image might have means of 60 and 30 counts with the same standard deviations. Its CNR is 30 / sqrt(16 + 16), or about 5.3. The darker image provides better separation in this example. Actual noise may increase as the signal falls, so measure it rather than assuming it stays constant.

For registered parallel and crossed images, a normalized difference can reduce sensitivity to a common multiplicative brightness change:

Subtract the measured dark offset first. Mask pixels whose denominator is too small, and keep exposure or gain differences calibrated. This quantity is not automatically a degree-of-polarization measurement and should not be labeled as one.

If an assembled display emits light during inspection, define its operating state. Display content, refresh timing, and existing polarization layers can change the measurement. Compare powered-off inspection with a controlled display state only where each represents a legitimate inspection condition.

Shortlist illumination parts without assuming polarization compatibility

Choose candidate lights by their documented geometry and connection requirements, then evaluate the complete polarized optical arrangement. Overall light dimensions do not establish the usable illuminated width at your working distance.

The LS-LSC-0527 is catalogued as a 300 x 17 mm line light source. Its suitability depends on the measured illumination profile, selected version, and ability to accommodate the required optical arrangement.

The LS-LSC-0441 is catalogued as a 493 x 60 mm oblique line light with natural cooling. Treat it as a candidate when an oblique illumination trial is justified by the sample response.

Neither description establishes an integrated polarizer, polarization purity, analyzer compatibility, or performance on a reflective screen. Request the relevant optical and mechanical information before treating either candidate as a production selection.

The application parts list should identify the camera, lens, selected light version, driver, source polarizer, analyzer, mounts, cables, and motion reference. Record which items are confirmed and which still require qualification. A light-only comparison is incomplete when the inspection depends on the full optical stack.

Validate the complete acquisition, not one favorable image

Use a fixed test matrix that changes one factor at a time. An initial study might include 3 illumination angles, 4 analyzer settings, and 2 exposure settings: 3 × 4 × 2 = 24 configurations.

First reject configurations with clipping, inadequate coverage, or unacceptable motion blur. Then rank the remaining configurations by feature separation across several part locations. Check the center, both edges, and any region where screen curvature or mounting changes the reflection angle.

For each promising configuration, retain:

• Raw images with the processing settings recorded.
• Feature and background measurements at defined locations.
• Detection decisions for acceptable and defective samples.
• Motion speed, exposure, gain, aperture, and encoder settings.
• Polarizer orientations and repeatable mechanical reference marks.
• Measurements after thermal settling and after a normal restart.

The lighting stability test provides a framework for checking drift. Extend the trial to cover different part lots and expected positioning variation, rather than repeatedly imaging one favorable sample.

If illumination is pulsed, confirm that the camera integrates during the useful optical output. Electrical trigger width alone does not establish the delivered exposure. The optical rise-time measurement guide explains that distinction.

Keep development samples separate from final verification samples. Tune the method on one set, then evaluate it on unseen acceptable and defective parts. Report false rejects and missed defects separately, with their sample counts. Zero misses in a small sample is not proof of zero production risk.

Recognize when polarization is the wrong answer

Crossed polarization can remove the reflection that carries the defect signal. A glossy scratch may become less visible when its specular contribution is suppressed. A different illumination angle may then outperform the polarized arrangement.

Polarization can also emphasize normal stress patterns, coating variation, adhesive structure, or display-layer behavior. These signals may look distinctive while having little relationship to the defect you need to reject.

Do not infer physical depth from a polarization-dependent color or brightness boundary. If the requirement is a step-height measurement, use a calibrated measurement method appropriate to the surface and required uncertainty. Polarization may help locate a boundary without measuring its height.

Finally, reconsider multiple-image analysis when motion registration is the dominant error. A stable single-orientation image can be a better production choice than a more elaborate polarization metric built from mismatched locations.

Check before ordering and release

Use this short checklist to close the application:

• Defect definition: Confirm the feature independently, rather than labeling it from image appearance alone.
• Coverage: Measure useful illumination across the entire camera line, including edge margins.
• Optical stack: Test the actual polarizers, lens, windows, and any diffuser together.
• Exposure budget: Confirm sufficient signal at the exposure required by production motion.
• Sampling: Check both cross-web resolution and travel-direction sampling for every acquired channel.
• Electrical compatibility: Verify the selected light version, driver, connectors, and cable requirements.
• Repeatability: Confirm that removal, refitting, restart, and normal part variation preserve the detection margin.
• Acceptance evidence: Retain images and decisions from independent verification samples.

Prepare the field width, minimum feature size, line rate, exposure, sample images, and polarizer arrangement before selecting parts. Start with the light configurator, or send those requirements through application support to review the illumination candidates and remaining compatibility checks.