Color camera near infrared imaging: IR-cut filter options

A color camera may record useful near-infrared images, but removing an IR-cut filter is not enough to establish that capability. You need a documented optical configuration, sufficient sensor response, suitable image processing and repeatable focus. This guide helps you decide whether to request manufacturer-authorized filter service, retain the camera unchanged or select a different camera. It also separates intentional near-infrared imaging from the unwanted infrared leakage that can corrupt ordinary color inspection.

Decide between documented conversion and a different camera

Start with the inspection requirement, not the filter retainer. Specify the illumination wavelength, smallest feature, maximum exposure time and whether visible color must remain accurate. A camera that detects infrared light may still fail your inspection because exposure is too long or material contrast is weak.

There are three sensible paths:

• Keep the existing optical stack when visible-color accuracy is the main requirement.
• Request a documented filter-service option when the manufacturer supports the intended configuration and can explain its limitations.
• Select a camera with documented response at the required wavelength when conversion is unsupported or performance remains uncertain.

Do not infer serviceability from hardware visible through the lens opening. An optical element may provide spectral filtering, environmental protection or both. The sensor package window is not an interchangeable accessory.

For the opposite problem—infrared entering an inspection that should measure visible color—use the guide to infrared contamination and its fixes. Its corrective goal differs from deliberate spectral conversion.

Understand what changing the optical stack actually changes

An IR-cut filter reduces infrared transmission while passing the intended visible band. Removing or replacing it changes more than the amount of light reaching the sensor. It can also change optical path length, reflections, calibration and protection against contamination.

A clear replacement window is not automatically equivalent to the original filter. Thickness, refractive index, coatings and mounting position all matter. Ask whether the supported service uses a replacement window or leaves a different documented optical arrangement.

The broader guide to choosing optical filters for contrast explains why passing more light is not always the best route to a more reliable inspection.

Do not treat a successful image immediately after service as acceptance. Establish image quality after warm-up, lens locking and repeated acquisition under the actual lighting conditions.

Check useful response at the illumination wavelength

Ask for spectral-response information for the exact camera configuration, including its window or filter. A sensor response curve alone does not describe transmission through the complete camera.

For a color camera, request channel-resolved response when available. Red, green and blue filter elements can transmit different amounts of near-infrared light. Their relative response can also change with wavelength. Visible color labels therefore do not tell you which pixels provide the strongest infrared signal.

A useful application test records:

• Exposure time, gain and aperture.
• Illumination wavelength and operating setting.
• Raw pixel format and any processing enabled.
• Target and background intensity distributions.
• Saturated pixels, dark offset and frame-to-frame variation.

For example, suppose your motion budget permits a maximum exposure of 0.8 ms. A bright image obtained at 8 ms does not prove suitability: it requires ten times the allowed exposure. Increasing gain may brighten the display without recovering the missing photon signal.

Test representative materials, not only a white reference card. Two visibly different surfaces can become similar under infrared illumination, while apparently similar materials can separate clearly.

Near-infrared reflectance imaging is also not a substitute for thermal imaging. A bright region under an infrared illuminator does not directly indicate elevated temperature. Follow the illuminator's documented optical-safety precautions; invisible brightness cannot be judged by looking at the source.

Account for the color-filter array and spatial sampling

A Bayer color camera places different color filters over neighboring pixels. Under near-infrared illumination, those filters may transmit unequal signals even when the target is uniform. Ordinary demosaicing can turn that imbalance into false color, texture or edge artifacts.

If the documented acquisition path provides raw mosaic data, preserve it during evaluation. Compare raw pixel groups before applying white balance, color correction, sharpening or compression. Those operations can obscure the cause of an image defect.

Do not assume that selecting one color channel preserves the full sensor sampling density. In a conventional Bayer pattern, red and blue samples occur at every other pixel position along each image axis. Interpolation creates additional output values, not additional measured detail.

Consider a hypothetical inspection with a 120 mm horizontal field of view and 2,400 sensor columns:

That feature spans four sensor-pixel pitches. If processing uses only red sample locations, the horizontal sample spacing is doubled, leaving approximately two sample intervals across the feature. Actual detectability also depends on alignment, contrast, lens performance and noise.

Use the resolution calculator for the initial geometric estimate, then validate the complete infrared processing path:

A monochrome camera avoids the Bayer mosaic, but it still needs suitable spectral response and optical transmission. Conversely, a color camera may remain adequate for large, high-contrast features when its raw data and processing are well characterized.

Recheck focus rather than relying on the visible setting

There are two separate focus effects. Changing a plate in the converging beam changes the optical path. Changing wavelength can also move the best-focus position because the lens may not correct visible and near-infrared light to the same plane.

For a plane-parallel plate near normal incidence, the approximate magnitude of its paraxial focus displacement is:

As a hypothetical example, a 1.0 mm plate with a refractive index of 1.5 produces an estimated displacement magnitude of:

This is an optical estimate, not a service adjustment value or the required travel of a lens focusing ring. Real assemblies need measurement, and replacing a plate requires the manufacturer's specified optical arrangement.

Focus using the intended illumination, aperture and working distance. Measure the center and corners, then repeat at the nearest and farthest target positions. A sharp center does not establish acceptable performance across the image.

Use a repeatable method from the guide to selecting an image sharpness metric. Keep exposure and illumination stable during the focus sweep so brightness changes do not masquerade as improved sharpness.

Compare camera candidates without assuming conversion support

The area scan cameras category and camera finder provide starting points for interface, shutter and format selection. Those characteristics do not establish infrared response or permission to alter the optical stack.

The MV-CU020-90GC is a color camera candidate for a configuration inquiry. The MV-CS200-10UM is a monochrome comparison candidate. Neither listing supplied here establishes infrared sensitivity, filter-service support or equivalent application performance.

All 2 share: Binning Supports 1 × 1, 2 × 2, 4 × 4; Decimation Supports 1 × 1, 2 × 2, 4 × 4; Lens mount C-mount; Humidity 20% RH to 95% RH (no condensation); Certification CE, RoHS, KC. Values as printed in each product's data sheet; open a part for its drawing and full specifications.

These are not a matched spectral comparison pair. Differences in sensor, shutter, interface and image sampling must be evaluated separately from the color-versus-monochrome question.

Before narrowing the choice, request the exact optical-stack description, spectral-response documentation and supported acquisition formats. If those answers are unavailable, do not base a production decision on a demonstration image alone. A camera specified for the intended wavelength gives you a more defensible starting point.

Choose the passband and verify mechanical fit

Removing infrared rejection admits a broader mixture of light; it does not isolate a useful infrared band. An external band-pass filter can help separate the intended illumination from visible light and other unwanted wavelengths.

The LS-LXR-0225 is listed as an 850 nm near-infrared band-pass filter with an M30.5 x 0.5 thread. Its listing does not establish bandwidth, peak transmission or out-of-band blocking, so obtain those details before selecting it for spectral isolation.

Check the filter thread on the lens itself. A camera's lens mount does not identify the lens's front accessory thread. Also verify clear aperture, available clearance and the effect of oblique rays on the selected filter's transmission.

Browse lens accessories or the accessories finder after confirming the mechanical interface. Do not force a filter into a similar-looking thread or assume an adapter cannot vignette the image.

Prove the complete configuration with an acceptance test

Write pass criteria before ordering service or replacement hardware. Otherwise, it is easy to accept a brighter picture without proving that the inspection improved.

1. Record the baseline. Save raw images and settings from the original configuration. Include acceptable parts, defects and difficult backgrounds.
2. Confirm the documented configuration. Record the camera variant, optical-stack service description, lens, filter and illumination arrangement.
3. Set the timing constraint. Use the exposure permitted by motion and cycle time, rather than allowing automatic exposure to hide a signal shortage.
4. Refocus under operating illumination. Check the full inspection region and expected target-height range before locking the lens.
5. Measure separation. Compare target and background distributions, not only their displayed appearance. Include repeated frames and multiple samples.
6. Repeat under disturbances. Change ambient lighting and target position within the defined operating envelope. Confirm that the decision remains stable.

For a simple screening example, suppose target and background means are 140 and 100 digital counts, with temporal standard deviations of 4 and 3 counts. Their difference is 40 counts, and the root-sum-square temporal variation is:

This is a limited temporal-noise check, not a guaranteed defect-detection probability. It excludes spatial texture, sample variation, drift and processing errors. The final acceptance test must use your actual decision algorithm and representative production variation.

Know when conversion is the wrong choice

Choose a different camera when the required exposure cannot produce adequate signal, the manufacturer does not support the optical change, or color-array artifacts erase the required detail. Replacement is also preferable when stable visible color and infrared measurement must operate together without a documented calibration strategy.

Alternating illumination can sometimes support separate measurements, but the acquisition and processing must keep those measurements distinct. The guide to sequencing multiple illumination wavelengths explains the timing approach; it does not establish that a particular camera supports both bands.

Before committing, check:

• The exact wavelength and material contrast are proven.
• The optical configuration is documented and supported.
• Spectral response is adequate within the exposure budget.
• Raw-data handling and any demosaicing are validated.
• Focus, filter fit and image cleanliness pass inspection.
• Acceptance criteria cover real sample and lighting variation.

Start a selection through the camera configurator, or send your wavelength, field of view, smallest feature and exposure limit through application support. Request confirmation of spectral suitability and supported service options before authorizing any optical-stack change.