| ISO 12233 Resolution | Horizontal spatial resolution | ISO 12233 slanted-edge or Siemens-star chart; record the response in line widths per picture height (LW/PH). | Approximately 1,200–2,000 LW/PH for a modern HD or 4K digital imaging module, depending on sensor format, optics, sharpening, and processing. | Higher LW/PH generally means finer target detail can be separated at the same distance. |
| SFR50 / MTF50 | ISO 12233 slanted-edge analysis at the center and image corners; report cycles per picture height or cycles per pixel. | A result near 0.20–0.35 cycles/pixel is a practical reference range for a well-focused long-range optical channel. | Measures contrast retention at medium spatial frequencies and helps identify focus, lens, or image-processing limitations. |
| Edge overshoot and sharpening | Inspect the ISO 12233 edge profile for ringing, overshoot, and excessive artificial sharpening. | Prefer controlled overshoot, normally below approximately 10% of the edge transition for measurement-oriented imaging. | Limits false detail and halo artifacts that can reduce reliable recognition of distant targets. |
| Signal-to-Noise Ratio | SNR at 10 lux | Uniform gray target, fixed exposure, fixed gain, stabilized temperature, and repeated frames; calculate 20 log10(signal/noise). | Approximately 30–40 dB for a low-noise visible-light channel with adequate exposure. | Higher SNR provides cleaner edges and more dependable target classification in low-light scenes. |
| SNR at 1 lux | Repeat the measurement at 1 lux while documenting exposure time, gain, frame rate, lens aperture, and infrared illumination status. | Approximately 18–30 dB is a useful comparison range for a sensitive digital camera system. | Shows how rapidly image quality degrades as ambient illumination falls. |
| Temporal noise stability | Capture at least 30 consecutive frames of a static target and calculate frame-to-frame luminance variation. | A stable system should maintain consistent noise behavior without visible fixed-pattern noise, banding, or gain pumping. | Indicates whether tracking, video analytics, and long-duration monitoring will remain reliable. |
| Low-Light Performance | Minimum usable illumination | Measure lux at the target plane with a calibrated illuminance meter; define “usable” as identifiable target shape with controlled noise and no excessive smearing. | Approximately 0.01–0.1 lux for sensitive monochrome or near-infrared-assisted operation; visible-color operation normally requires more light. | Provides a more meaningful comparison than an unspecified “night vision” label. |
| Near-infrared response | Test with a documented 850 nm or 940 nm illuminator, recording optical power, beam angle, distance, and camera gain. | Report image SNR and target contrast separately for 850 nm and 940 nm operation. | Prevents misleading comparisons caused by different infrared wavelengths and illumination power. |
| Dynamic range | Capture a calibrated step chart under high-contrast lighting and report the usable exposure range before highlight clipping or shadow loss. | Approximately 60–80 dB can be used as a broad engineering reference for modern industrial imaging modules. | Higher dynamic range helps preserve detail in scenes containing bright sky, headlights, and dark foreground areas. |
| Detection and Recognition Range | Detection range | Use a documented target size, focal length, sensor pixel pitch, atmospheric visibility, and illumination level; detection is commonly associated with roughly 2 pixels across a critical target dimension under Johnson-style criteria. | For a 1.7 m human-sized target, a 400–800 mm equivalent optical channel may provide a calculated detection range of roughly 1–4 km in clear conditions. | Detection means noticing a target is present; it does not confirm identity or fine details. |
| Recognition range | Use the same target and atmospheric conditions, but require substantially more pixels across the target’s critical dimension, commonly around 8 pixels for recognition-oriented testing. | A practical range may be approximately 0.3–1.5 km for a 1.7 m human-sized target, depending strongly on optics and image quality. | Recognition requires more spatial detail than simple detection and should never be inferred from detection range alone. |
| Identification range | Require a larger pixel density on the target, controlled scene geometry, and independent human evaluation or a validated image-analysis protocol. | Often limited to approximately 0.1–0.7 km for human-sized targets in practical outdoor conditions. | Identification is affected by atmospheric turbulence, motion blur, compression, focus accuracy, and target orientation. |
| Long-Range System Validation | Optical focal length and field of view | Record actual focal length, sensor dimensions, horizontal field of view, and optical zoom position for every measurement. | A narrow field of view improves pixel density on distant targets but reduces situational coverage. | Ensures detection-range claims are linked to a clearly defined optical configuration. |
| Atmospheric visibility | Document visibility distance, humidity, rain, haze, temperature, and wind during outdoor testing. | Use clear-air results and degraded-weather results as separate data sets; do not combine them into one range figure. | Atmospheric scattering can reduce contrast before geometric resolution becomes the limiting factor. |
| Data reporting completeness | Require raw test images, test charts, exposure settings, gain, frame rate, firmware version, calibration date, and measurement uncertainty. | A technically credible report should disclose test conditions and repeatability, not only a maximum-distance number. | Improves supplier comparability and reduces performance claims that cannot be independently reproduced. |
