Axis Communications’ white paper, Thermal cameras, explains that thermal sensors create images from infrared radiation emitted by objects as a function of temperature. They do not depend on visible illumination. Small temperature differences can make a person or vehicle stand out against darkness, shadows, camouflage, or a visually complex background.
\nThat strength is primarily detection and shape recognition. Axis states that thermal images alone are generally insufficient to identify a specific person. In an integrated perimeter workflow, a thermal camera can detect an intruder and direct a visual camera toward the event so an operator can verify and, where the installed view supports it, identify the subject.
\nThermal performance should not be reduced to a single attractive range figure. Focal length, sensor resolution, target dimensions, required task, temperature difference, analytics requirements, and environment all matter. Axis discusses Johnson’s criteria as a planning method and notes that analytics may require more pixels than the human-observer rules of thumb.
\n\nThermal imaging often performs better than visual imaging in darkness, moderate fog, smoke, or difficult lighting, but it is not weather-proof vision. Rain, fog, and smog attenuate thermal radiation and reduce detection range. Weather and solar heating can also make target and background temperatures more alike. The source’s planning examples assume a temperature difference; actual site performance can differ.
\nNoise equivalent temperature difference, or NETD, describes a sensor’s ability to distinguish small thermal differences, with a lower value generally indicating better sensitivity. Axis cautions against rating cameras only by NETD because measurement methods are not standardized sufficiently for simple cross-product comparison. Complete camera processing and the real scene affect the delivered image.
\n\nWrite the thermal requirement as a detection statement: target class, minimum size, direction, speed, near and far boundaries, environmental range, alarm zones, and maximum acceptable nuisance rate. Separately define what the visual camera must provide after an alarm—overview, recognition, identification, color, clothing detail, plate capture, or simply confirmation that a person is present.
\nDo not imply that a long thermal detection range is also an identification range. Show both boundaries on the site plan. If the visual camera cannot cover the thermal alarm area with the required detail and lighting, the response chain has an evidence gap.
\n\nKeep a thermal clip and corresponding visual clip for each critical test point, along with lens, firmware, analytic version, palette, zone, time, weather, target, result, and reviewer. Record the longest verified detection distance, not merely the manufacturer’s calculated maximum.
\nRevalidate after foliage changes, construction, new fencing, camera movement, lens or firmware changes, analytic updates, or repeated nuisance alarms. The successful system is not a thermal camera with an impressive picture. It is a measured detection layer connected to visual verification and a response procedure that operators have demonstrated under realistic conditions.
\n\n