Hangzhou City, Zhejiang Province, China-October 15, 2026

Why Infrared Filters Are Indispensable in Autonomous Driving Night Vision Systems？

At Night, Visible-Light Cameras Are Nearly Useless. For Autonomous Driving Systems to Keep Working After Dark, the Perception Stack Needs a Channel That Doesn’t Depend on Visible Light.

That channel is infrared night vision. And whether infrared night vision can truly be deployed depends heavily on a seemingly inconspicuous optical component mounted at the very front of the system: the long-wave infrared filter.

1. Why Infrared Is Needed at Night

Visible-light cameras perform well during the day and in well-lit environments—high resolution, accurate color reproduction, and low cost. But at night, in backlight, heavy fog, or rainstorms, they fall short. An autonomous driving system cannot simply shut down because the weather is bad. Humans drive at night relying on headlights, experience, and anticipation of road conditions. Machines must rely on sensors.

Millimeter-wave radar can penetrate rain and fog, but its resolution is insufficient to distinguish whether the object ahead is a person or a plastic bag. LiDAR offers high precision, but suffers severe attenuation in heavy rain and snow—and its cost remains high.

Infrared thermal imaging does not rely on visible light. All objects above absolute zero radiate infrared energy. People, animals, vehicles, and road surfaces each have distinct thermal radiation signatures. Thermal imagers capture the temperature difference information from this radiation and can operate day or night in most scenarios. Heavy fog blocks visible light, but long-wave infrared at 8–14 μm experiences relatively low attenuation.

Infrared is not omnipotent, however. During high-noon heat, surface thermal contrast decreases, and infrared imaging performance suffers. Detection range under rain and fog also varies with weather conditions. Its role is to provide a supplementary channel when existing sensors fail—not to replace them.

2. Four Nighttime Driving Scenarios Place Different Demands on Filters

Breaking nighttime driving into distinct scenarios reveals that different situations impose different requirements on filters. Lumping them together makes it easy to discuss only generic specifications.

Nighttime driving on unlit roads is the most common scenario. Here, road surface temperature is close to the surrounding environment, and the thermal contrast between people/vehicles comes primarily from the temperature difference between areas illuminated by headlights and the human body or engine compartment. The most critical requirement for the filter in this scenario is out-of-band rejection. Streetlights, vehicle lights, and oncoming headlights are all strong stray light sources. If out-of-band blocking is not deep enough, stray light will directly overwhelm the valid signal.

Backlight and exiting a tunnel is the second scenario. Visible light becomes overexposed or even blinding at that moment, while the infrared channel remains unaffected. But the filter must ensure no lag response during such drastic lighting changes—otherwise, the image’s light-to-dark transition cannot keep up with vehicle speed, and frames will be dropped.

Heavy fog and rainstorms is the third scenario. Long-wave infrared attenuates less under these conditions, but water vapor has strong absorption in specific bands. Band selection must avoid absorption lines; otherwise, even the best filter cannot transmit a complete signal. Chalcogenide glass offers good transmission across the 2–12 μm range, providing wider selection space in the 8–12 μm band than germanium substrates.

Direct intense light is the fourth scenario—and the one with the strongest stray-light impact. Oncoming headlights, specular road reflections, and tunnel lighting all create very high irradiance on the front surface of the filter. This scenario places the highest demands on the coating’s anti-reflection and out-of-band rejection performance. Poor coating adhesion or insufficient blocking will directly manifest as grayish images and blurred boundaries of thermal targets.

3. What Exactly Does the Filter Block in the System?

Many people assume a thermal imager is just a “camera that sees heat”—install it and it works. In reality, it is not that simple.

The role of the long-wave infrared filter is to transmit only thermal radiation in the 8–14 μm band while blocking visible light, near-infrared, and mid-wave infrared stray light as much as possible. Sunlight during the day contains not only visible light but also mid-wave and far-infrared components. If out-of-band blocking is insufficient, these stray light components mix into the receiver, significantly reducing imaging contrast. In addition, background radiation in the mid-wave infrared band itself can interfere with long-wave infrared receivers.

The filter’s out-of-band rejection capability directly affects signal-to-noise ratio and temperature measurement accuracy. According to the Lambert-Beer law, A = lg(I₀/I) = εcl. Additional reflection introduced by the coating effectively raises the signal baseline, and in quantitative measurement scenarios, this error translates directly into deviations in concentration or temperature.

A long-pass coating is essential. Substrate materials typically include germanium (Ge), zinc selenide (ZnSe), or chalcogenide glass—all offering high transmittance and good chemical stability in the long-wave infrared band. Take germanium as an example: its refractive index is about 4.0, and single-interface reflection loss in the 8–14 μm band is approximately 36%. Without a long-pass coating, optical energy loss would be too high.

Coating design must also consider thermal expansion coefficient matching. If the thermal expansion difference between substrate and coating is too large, repeated thermal cycling will cause film cracking or delamination. This is not a problem that one-time sampling inspection can detect—it accumulates continuously over the vehicle’s entire lifecycle.

4. Automotive-Grade Reliability Is Not a Bonus—It’s a Threshold

Consumer-grade products can be replaced if they fail. Automotive-grade products can compromise safety if they fail. Automotive filters must withstand temperatures from –40°C to +85°C, as well as vibration, humidity, salt spray, and UV aging. Out-of-band blocking depth must be pushed to OD4 or higher, and salt spray resistance and adhesion also have clear thresholds.

Each of these specifications corresponds to a failure consequence: insufficient blocking allows mid-wave background radiation to enter; delamination causes blistering in humid environments; failing salt spray leads to coating corrosion in coastal conditions. These are not lab bonuses—they are the threshold for getting onto a vehicle.

More importantly, failure is gradual. Film degradation, cracking, and transmittance decay do not manifest immediately—they cause the entire night vision system’s performance to drift slowly over time. By the time it is visible to the naked eye, the degradation has accumulated to an irreversible degree.

MULTI IR has a relatively complete product portfolio in automotive-grade infrared filters. Our recently exhibited chalcogenide glass substrate with DLC (diamond-like carbon) coating solution operates in the 8–12 μm band, with average transmittance no less than 93%, coating hardness greater than 18 GPa, cross-hatch adhesion grade 5B, and salt spray resistance exceeding 1000 hours. It has already been deployed in volume in automotive infrared applications.

For automotive scenarios, we conduct automotive-grade reliability validation including high-low temperature cycling and damp-heat aging. Such tests are typically not mandatory for consumer-grade products, but they are a basic threshold for automotive-grade applications.

5. Fusion Calibration: Filter Problems Get Amplified

Infrared night vision does not work in isolation. It must fuse data with visible-light cameras, millimeter-wave radar, and LiDAR. Fusion requires calibration, and calibration requires that each sensor’s output data be sufficiently stable and trustworthy.

If the front-end filter performance is unstable—for example, transmittance drifting with temperature, or insufficient out-of-band rejection causing SNR fluctuations—then the temperature data and image data output by the infrared channel will have deviations. These deviations may not be obvious when viewed in isolation, but once they enter the fusion stage, they can be amplified.

For example: LiDAR detects an obstacle 150 meters ahead, the thermal imager detects a heat source at the same position, and the visible-light camera captures a blurry outline. The data from three sensors must be fused into one target. If the infrared channel has a deviation due to a filter issue, the fusion algorithm may identify this target as two separate objects—or miss it entirely.

The perception quality of a night vision system is determined by the data quality at the starting point of the optical chain, and there is no second remedial step on that chain. The filter is the front-most optical component in an infrared night vision system, and its quality determines the starting point of the entire chain. If the starting point has problems, the reliability of the entire downstream chain is compromised.

MULTI IR – National-level “Little Giant” Specialized and Sophisticated Enterprise. Over 10,000 types of infrared sensitive components in stock. Top 3 globally in comprehensive strength. Product portfolio covers infrared filters, optical coatings, infrared sensors, and other core categories. Lead drafter of the infrared filter industry standard.

Website: www.mirhz.com | Global Site: www.miroptech.com

About us

Founded in 2007, Hangzhou MULTI IR Technology Co., Ltd. is an optoelectronic technology enterprise integrating R&D, production, and sales. Its products are widely applied in aerospace, medical care, AR/VR, display imaging, photography, and other fields, steadily holding the position of the world’s largest spot supplier of optical components.

Media Contact
Company Name: HANGZHOU MULTI IR TECHNOLOGY CO., LTD.
Contact Person: Media Relations
Email: Send Email
Country: China
Website: https://www.miroptech.com/