IR and night vision: how cameras see in the dark

The invisible spectrum: what the human eye cannot see

When we think of total darkness, we assume nothing can be recorded without powerful floodlights. However, visible light represents only a tiny fraction of all electromagnetic radiation traveling through the universe. Just above red in our perception spectrum lies the infrared band, an energy invisible to our eyes but perfectly detectable by specialized electronic components. In practice, this means we can illuminate an environment entirely in the dark using light-emitting diodes that generate no visible glare for humans, keeping surveillance completely discreet.

To understand how this works in practice, imagine a standard light bulb. It emits radiation across multiple frequencies simultaneously, generating white light. Infrared (IR) emitters, on the other hand, concentrate their energy at specific wavelengths, usually measured in nanometers, such as 850nm or 940nm. 850nm LEDs emit a faint reddish glow when observed up close, but reach longer distances. 940nm emitters are completely invisible, though they reach shorter distances. This design choice directly depends on the balance between operational stealth and desired illumination range in a security project.

The anatomy of the sensor: how chips turn light into images

Capturing infrared radiation requires more than just an ordinary lens; it requires semiconductors with high photosensitive sensitivity. The vast majority of modern cameras use CMOS sensors (complementary metal-oxide-semiconductor), tiny chips packed with millions of microscopic dots called pixels. Each pixel acts as a photon collector, converting incoming light into a measurable electrical charge. During the day, these sensors process the full spectrum, but they need a special filter to prevent false color distortions caused by excess infrared rays.

This critical component is the ICR filter (infrared cut filter). During the day, a small mechanical motor positions this glass filter between the lens and the sensor, blocking infrared light and ensuring true-to-life colors. As soon as ambient light drops below a predetermined threshold, a photoresistor alerts the system. The motor physically removes the filter, allowing both visible light and infrared to reach the sensor simultaneously. In practice, the image loses color and becomes monochrome in grayscale, but gains a massive capability to see in the dark.

Active illumination versus low-light amplification

There are basically two distinct technological approaches to seeing in the dark: active illumination and low-light amplification. The first, most common in commercial security cameras, uses built-in infrared LEDs to bathe the scene in invisible radiation, acting like a flashlight only the camera can see. This approach guarantees clear images even in absolute pitch black, but suffers from physical limitations such as excessive reflection on bright surfaces or light scattering in dense fog.

On the other hand, low-light amplification — widely used in military night vision goggles — works by collecting and multiplying the few photons available in the scene. Even on a moonless night, diffuse sources of natural light exist, such as starlight and residual airglow. An image intensifier tube captures this faint light, causes electrons to collide with a microchannel plate, and generates a glowing greenish image on a phosphor screen. While it requires no dedicated illuminators, this technology loses effectiveness in absolute darkness and requires extremely expensive optical components to prevent severe distortion.

Choosing between these methodologies strictly depends on the use case and available budget. While residential and commercial systems rely on the robustness and low cost of active infrared LEDs, rescue applications and long-range monitoring frequently demand extremely sensitive sensors capable of operating under starlight without alerting observers via visible light beams.

Thermography: seeing heat instead of light

There is yet an entirely different category of night vision that ignores light altogether: thermography. Instead of recording photons reflected by surfaces, thermal cameras capture long-wave infrared radiation, which is simply heat emitted by bodies, animals, and objects. All elements with temperatures above absolute zero radiate thermal energy proportional to their physical state. In practice, this means a person will stand out clearly on a dark, cold night not because light is shining on them, but because their body temperature contrasts sharply with the surrounding environment.

Thermal cameras use microbolometer arrays, tiny sensors that change their electrical resistance upon absorbing radiated heat. An internal processor maps these electrical variations and converts them into an artificial color palette, where warm tones typically appear in white or yellow and cold tones in blue or black. This technology is immune to visual tricks such as camouflage, dense smoke, fog, or total darkness. However, it comes with a significantly higher manufacturing cost and does not provide sharp details of faces or textures, focusing essentially on presence detection and thermal variations.

Physical limitations and common project pitfalls

Implementing night vision systems requires attention to physical details that frequently catch engineers and installers off guard. One of the most recurring problems is backscatter reflection, which occurs when infrared light hits nearby reflective surfaces head-on, such as white walls or glass, temporarily blinding the camera sensor. Another critical point is insect attraction: because infrared LEDs operate at frequencies close to visible for some animals, they frequently attract moths and mosquitoes close to the lens, generating false alarms in motion detection systems.

Additionally, the internal thermal management of the camera itself becomes a challenge during prolonged night operations. Image processing circuits and the LED emitters themselves generate considerable heat. If the equipment housing does not dissipate this energy properly, thermal saturation of the CMOS sensor occurs, resulting in excessive digital noise — those static granulations that ruin image sharpness precisely when analytical clarity is most needed in critical environments.

Final thoughts on the future of night vision

The continuous evolution of image sensors promises to radically transform how we approach surveillance in dark environments. New backside-illuminated sensor technologies combined with advanced artificial intelligence algorithms can already restore colors and details in environments with near-imperceptible illumination, reducing the dependency on bulky infrared floodlights. Understanding the physics behind these lenses allows us to design more efficient, robust security systems tailored to the real operational needs of the physical world.