Bandpass Filters
Bandpass Filters in Machine Vision: A Practical Guide
What Is a Bandpass Filter?
A bandpass filter is an optical filter that allows only a specific range of wavelengths to pass through while blocking everything outside that range. Think of it as a precise "window" that lets in just the light you want your camera to see – nothing more, nothing less.
Every bandpass filter is defined by two key specifications:
- Center Wavelength (CWL) – The midpoint of the wavelengths the filter transmits. For example, a filter with a 530nm center wavelength is optimized for green light.
- Bandwidth (FWHM) – The width of the transmission window, measured as "Full Width at Half Maximum." A 530nm filter with a 30nm bandwidth transmits light from roughly 515nm to 545nm.
In machine vision, this selectivity is powerful. By controlling exactly which wavelengths reach your sensor, you can dramatically improve contrast, eliminate interference from ambient light, and make defects or features "pop" that would otherwise blend into the background.
Why Use a Bandpass Filter in Machine Vision?
Machine vision systems perform best when the feature you're inspecting stands out clearly from its surroundings. In an ideal world, your target would be bright and your background would be dark (or vice versa). Bandpass filters help you get closer to that ideal by giving you control over what your camera actually "sees."
Here's what a well-chosen bandpass filter can do for your system:
Match your illumination for maximum efficiency
When you pair a bandpass filter with a specific LED wavelength – say, a 660nm red LED with a 660nm bandpass filter – your camera captures only the light you're actively providing. Ambient light from windows, overhead fixtures, or adjacent workstations gets blocked. The result is consistent imaging regardless of environmental changes.
Enhance contrast between materials
Different materials reflect and absorb wavelengths differently. A red apple and a green apple look similar under white light, but under red illumination with a red bandpass filter, one appears bright and the other dark. The same principle applies to industrial inspection: choosing the right wavelength can make surface defects, contamination, coatings, or printed markings far more visible than they would be under broadband lighting.
Reduce noise and simplify processing
By blocking unwanted wavelengths, you're effectively removing visual clutter before it ever reaches your sensor. This means cleaner images, better signal-to-noise ratios, and often simpler – and faster – image processing algorithms downstream. In some cases, a good optical filter can eliminate the need for complex software corrections entirely.
Improve consistency across production runs
Ambient light changes throughout the day. Seasons change. Bulbs age and shift in color. A bandpass filter paired with a stable LED light source isolates your imaging system from all of these variables, giving you repeatable results shift after shift, day after day.
Common Applications for Bandpass Filters
Presence/absence detection
One of the simplest but most valuable applications. A narrowband filter matched to your light source ensures your sensor responds only to items illuminated by your system – not to reflections, shadows, or stray light. This is especially useful on fast-moving production lines where split-second decisions matter.
Surface inspection and defect detection
Scratches, dents, stains, or coating inconsistencies often become visible only under specific wavelengths. For example, organic contamination on a metal surface might fluoresce under UV light – a UV-passing bandpass filter paired with a UV LED can make invisible residues glow brightly against a dark background. Similarly, shallow scratches on polished surfaces may only show contrast under particular angles and wavelengths.
Code and character reading (OCR/OCV)
Whether you're reading laser-etched serial numbers on metal, inkjet-printed lot codes on cardboard, or embossed date stamps on plastic, the right bandpass filter improves edge definition and character contrast. This is particularly valuable when the marking and background are similar in color but respond differently to specific wavelengths.
Color sorting and differentiation
When you need to distinguish between items that look nearly identical under white light, a bandpass filter tuned to the wavelength where they differ most can simplify your system dramatically. Instead of requiring an expensive color camera and complex processing, a monochrome camera with the right filter can make a binary bright/dark distinction – faster, cheaper, and more reliable.
Fluorescence imaging
Some materials naturally fluoresce (emit light at a different wavelength) when excited by UV or specific visible wavelengths. Bandpass filters are essential here: one filter on your light source to deliver only the excitation wavelength, and another on your camera to capture only the emitted fluorescence while blocking the excitation light. This technique is widely used in detecting adhesives, lubricants, coatings, and biological contamination.
Laser profiling and 3D measurement
In laser triangulation systems, a bandpass filter centered on your laser's wavelength (commonly 635nm, 660nm, or 450nm) ensures the camera sees only the laser line – not ambient light or reflections from other sources. This is critical for accurate height measurement and 3D surface reconstruction.
How to Choose the Right Bandpass Filter
Step 1: Start with your light source
Your filter should match – or be very close to – the wavelength of your illumination. If you're using an 850nm infrared LED, start with an 850nm bandpass filter. If you're using a 470nm blue LED, look at filters centered around 470nm. This pairing maximizes the light your camera receives while rejecting everything else.
Step 2: Decide on bandwidth
- Narrow bandwidth (10–25nm) offers maximum selectivity. It's ideal when you need to isolate a very specific wavelength, such as in fluorescence or laser applications. The trade-off: less light reaches your sensor, so you may need brighter illumination or longer exposure times.
- Medium bandwidth (30–50nm) balances selectivity with light throughput. This is a good general-purpose choice for most inspection tasks.
- Wide bandwidth (60–100nm) lets in more light and is more forgiving of slight wavelength mismatches. Useful when you need speed and brightness but don't require razor-sharp spectral isolation.
Step 3: Consider your environment
High ambient light or mixed lighting conditions (like a factory floor with skylights and fluorescent overheads) call for tighter filtering. A well-controlled enclosure with dedicated lighting may allow you to use broader filters and faster exposures.
Step 4: Test before you commit
Whenever possible, image your part under different single-wavelength LEDs first. The wavelength that gives you the best natural contrast between your feature of interest and the background is likely your ideal filter center wavelength. Once you've identified that, selecting the right bandpass filter becomes straightforward.
A Few Practical Tips
- Don't forget about filter mounting. Make sure your filter size matches your lens thread or housing. Common machine vision sizes include M25.5, M27, M30.5, and M35.5 threads, as well as various drop-in and slip-on mounts.
- Check optical density (OD) for blocking. For demanding applications, especially fluorescence, you want high OD (3.0 or greater) outside the passband to ensure stray light is truly eliminated – not just reduced.
- Consider anti-reflection coatings. A good AR coating reduces surface reflections that can cause ghosting or reduce transmission efficiency.
- Mind the temperature. Some filters shift their center wavelength slightly at extreme temperatures. If you're operating in a high-heat environment (near furnaces, ovens, or in direct sunlight), ask about thermal stability.
Bringing It Together
A bandpass filter is one of the most cost-effective upgrades you can make to a machine vision system. For a relatively small investment, you gain control over your imaging environment, improve contrast, reduce processing complexity, and achieve more consistent results.
The key is matching your filter to your light source and application. When those elements align, your camera sees exactly what it needs to see – and nothing it doesn't.
Ready to find the right bandpass filter for your application? [Explore our Bandpass Filter range → https://www.kupooptics.com/en/collections/bandpass-filters] or contact us for application support.
Frequently Asked Questions
https://www.kupooptics.com/en/blogs/filter-products/mv_bandpass_filter
What Is a Bandpass Filter?
A bandpass filter is an optical filter that transmits a defined range of wavelengths (the "pass band") while blocking wavelengths outside that range. It is the most selective standard filter type in machine vision: rather than passing everything above or below a cutoff wavelength, it passes only a controlled window of the spectrum. The pass band is defined by two parameters: center wavelength (CWL) — the midpoint of the transmission window; full-width at half-maximum (FWHM) — the width of the window measured at 50% of peak transmission.
Why Use a Bandpass Filter in Machine Vision?
Bandpass filters are used when spectral selectivity is critical. Common use cases: matching a bandpass filter to a specific LED illumination wavelength (e.g., 850 nm NIR LED + 850/30 nm bandpass filter) to reject all ambient light except the illumination band; separating multiple spectral channels in multispectral or hyperspectral imaging; isolating fluorescence emission from excitation in fluorescence imaging; eliminating unwanted spectral overlap in color sorting systems; reducing the effect of surface color variation when only structural information is needed.
What Are the Key Specifications of a Bandpass Filter?
Center Wavelength (CWL): the nominal midpoint of the pass band. Full-Width at Half-Maximum (FWHM): the width of the pass band at 50% transmission — determines spectral selectivity. Peak transmission: how much light in the pass band actually passes through — should exceed 80% for most applications. Out-of-band blocking: optical density (OD) outside the pass band — OD ≥ 3 is standard; OD ≥ 5 for fluorescence or laser work. Angle sensitivity: CWL blue-shifts at non-zero angles of incidence — critical for non-telecentric systems.
How Do You Match a Bandpass Filter to Your Illumination?
The goal is to maximize the transmission of your illumination wavelength while blocking everything else. Steps: identify the peak emission wavelength of your light source (from the LED datasheet — it's usually specified as the dominant or peak wavelength). Choose a bandpass filter with a CWL that matches this peak (±5 nm is usually acceptable). Choose an FWHM that is wider than the spectral width of the LED (LEDs typically have a FWHM of 20–40 nm for standard types, 1–5 nm for laser diodes). Confirm that the filter's blocking level (OD) in the ambient light range is sufficient to suppress the background you expect.
What Is the Difference Between a Narrow and Wide Bandpass Filter?
Narrow bandpass filters (FWHM ≤ 10 nm): high spectral selectivity, used in laser line isolation, fluorescence, and spectroscopy; lower peak transmission due to tighter coatings. Wide bandpass filters (FWHM 50–100 nm): used when spectral channel separation is needed but not laser-level precision; higher transmission, easier to manufacture, lower cost. For most machine vision illumination-matching applications, bandpass filters with FWHM of 25–50 nm offer the best balance of selectivity and throughput.
How Does Angle of Incidence Affect a Bandpass Filter?
At non-zero angles of incidence, the CWL of a bandpass filter shifts toward shorter wavelengths (blue-shift). The shift magnitude depends on the filter design (Fabry–Pérot or multi-cavity) and the angle. For thin-film interference bandpass filters, the CWL shift is approximately: Δλ ≈ -λ₀ × sin²(θ) / (2n²), where θ is the angle and n is the effective refractive index of the coating stack (~1.45 for standard coatings). At 10°, the shift is typically 1–3 nm; at 20°, it can reach 5–10 nm. This matters most for narrow bandpass filters (FWHM ≤ 10 nm) in non-telecentric optical systems.
