Infrared and Thermal Cameras
“Infrared camera” can refer to different technologies. Near-infrared systems observe reflected illumination close to visible light, while thermal cameras measure longer-wave radiation associated with surface temperature.
Near-infrared and thermal imaging
Near-infrared cameras often work with an infrared illuminator and can support driver monitoring or night imaging without visible glare. Their images still depend on reflected energy.
Thermal cameras, commonly using long-wave infrared, measure radiance in a wavelength band where warm people, animals and machinery can contrast with the background. They do not directly measure a universal object temperature, and they do not see through every obscurant or material.
Automotive roles
Exterior thermal imaging can support:
- night-vision displays for the driver;
- pedestrian and animal detection in low visible light;
- an additional perception channel for automatic emergency braking research;
- detection of warm machinery or road users whose visible contrast is weak.
Near-infrared interior cameras can support gaze, head-pose and eyelid estimation across changing cabin light. The vehicle's exact wavelength, illumination and algorithm should be stated rather than grouping all infrared systems together.
Strengths
Thermal sensing does not require visible headlights to create contrast, so a warm pedestrian can remain detectable in darkness or against some visually confusing backgrounds. This can complement the color and texture of a visible camera.
Research has evaluated thermal imaging for pedestrian-protection scenarios, and earlier work has examined fusion of visible and thermal images. See Transportation Research Record: thermal cameras for pedestrian protection and Sensors: visible and thermal image fusion for pedestrian detection. These studies support technical potential; they do not prove the performance of every production vehicle.
Limitations
Thermal images usually have less fine texture and no visible color information for reading a traffic light or painted sign. Performance can change when target and background temperatures are similar, when a warm road or engine dominates the scene, or when rain, fog, snow and atmospheric absorption reduce contrast.
Common automotive glass is not transparent across all thermal-infrared wavelengths, so sensor placement and protective-window materials matter. Dirt, water and ice on the optical surface still obstruct the measurement. Resolution, field of view, frame rate, thermal sensitivity and calibration all affect useful detection.
Manufacturer range claims also depend on target type, weather, detection algorithm and whether the result is one thermal contrast, a classified pedestrian or a stable track. FLIR describes automotive thermal-camera capabilities at FLIR Automotive Development Kit; these are manufacturer claims that require vehicle-level validation.
Fusion with other sensors
Thermal and visible images can add different evidence: thermal contrast for warm road users, visible texture and color for scene semantics. Radar can add range and radial velocity. Fusion still requires alignment, synchronization, confidence management and training data representing the joint failure modes. See Sensor fusion.
An infrared sensor is therefore not a universal cure for darkness or weather. It is another measurement channel whose value depends on the assigned function and operational design domain.
How to evaluate a system
Ask whether the claim concerns display enhancement, object detection, warning or automatic intervention. Tests should disclose target temperature, background, range, weather, lens condition, false positives, software version and full vehicle response.
The distinction between sensor measurement, perception and braking behavior is covered in How to validate sensor and perception claims. A successful thermal image is not by itself proof that the vehicle will classify the object, predict its path and stop safely.