Optical ALPR evidence — XIAO 0.6.0

Flock Noir measures light changes and combines them with nearby ALPR radio candidates. Its current 8–12 Hz profile is a test configuration. This project has no verified basis for claiming that every ALPR, or every Flock camera, emits 10 Hz / 20 ms flashes. Illumination and exposure depend on camera design and configuration; see Axis's plate-capture guide. A matching waveform is supporting evidence, never a camera identity check.

What the sensors measure

OPT101 integrates a photodiode and amplifier. Powered from a compatible 3.3 V breakout, OUT feeds D1 / GPIO2, ADC1. The device responds to visible and near-infrared light; it cannot measure wavelength. Sampling at ~1000 samples/s permits millisecond pulse timing. The firmware removes a slowly changing baseline, uses a noise-dependent threshold with hysteresis, and requires:

These constants are in firmware/FlockNoir/config.h. Profiles outside these bounds can be missed, including continuous illumination. Calibrating a different profile requires known-source measurements and new negative controls, not an assumption that a new frequency identifies a manufacturer.

OV2640 supplies spatial brightness evidence. It captures 640×480 JPEG at fixed exposure/gain with two PSRAM buffers. A separate task extracts 80×60 block brightness, bright-area fraction and centroid, retaining the full-quality JPEG for preview. The camera searches for localized repeated brightness changes; whole-frame flicker and very weak changes are rejected.

At the reference ~25 fps, frames are ~40 ms apart. A 20 ms pulse can fall between frames. In particular a 10 Hz source may appear to blink at 5 Hz. The detector keeps this as an explicitly aliased candidate, halves its confidence score and records the observed frequency. It does not invent an exact pulse width or call the observed rate 10 Hz. The CSV field camera_timing distinguishes aliased_candidate from frame_limited_pattern. Even the latter has limited timing precision. Use OPT101 for the direct timing measurement.

Combined evidence

Only ALPR mode runs both optical paths, together with Flock-You WiFi/OUI/probe and BLE rules. Each source has its own timestamp; fresh evidence within 3 seconds can produce ir+camera+ble+wifi or a subset in the ALPR CSV. Correlation works in either arrival order. Shared vendor prefixes remain weak hints; a nearby radio and a light source are not assumed to be the same object.

Pig Detector, Wardrive and General Scanner pause optical analysis. Their mode switches clear stale optical results so a previous signal cannot trigger a new mode's buzzer. Camera preview is a separate view, not an implicit mode change.

Bench validation before field interpretation

1. Use the wiring guide. Select ALPR and enable OPT101 only after wiring. Cover/uncover the sensor: raw counts and waveform should change. ADC values alone cannot establish a physical connection.
2. Check the reported OPT101 sample rate (~1 kHz), camera analysis rate (~25 fps), clipping and sample-gap counters under actual lighting. Saturated OPT101 output can be below ADC full-scale: a flattened waveform still needs attention.
3. Aim OPT101 and camera at a separate, measured 10 Hz / 20 ms light source. OPT101 should report about 10 Hz, 20% duty and 20 ms width after four valid intervals. A 25 fps camera can legitimately report a 5 Hz alias candidate.
4. Negative controls: cover the sensors, constant light, 50/60 Hz light, 20 Hz pulses, very short / long pulse widths, irregular pulses and saturation. These must not produce an OPT101 timing match. A TV remote checks sensitivity; it is not a valid substitute for the reference waveform.
5. Inspect ALPR CSV source/method and pulse fields. Add a known radio candidate only when testing correlation; observe that a camera-alone event does not invent a MAC or radio tier. Missing SD/GPS must remain visible.
6. Switch to Pig Detector and Wardrive. OPT101 sample rate / camera analysis should pause, the prior alert should clear, and ALPR logging should stop. Restore ALPR and verify both optical paths resume without rewiring GPS.

Synthetic tests cover waveform timing and all phases of the 25 fps alias example. Hardware smoke checks measure execution rates and mode isolation. Neither proves range, sensitivity or false-positive rates against real ALPRs. Measure and record actual target waveforms before making manufacturer-specific claims.

Field setup

Use an OV2640 without its IR-cut filter. Fix focus and aim the camera and OPT101 at the same scene. Optical shade or a characterized filter can reduce ambient light; neither identifies the source. Use the built-in fixed exposure/gain as a starting point, then evaluate waveform headroom outdoors. Nearby light may come from ordinary security cameras, traffic hardware or other emitters.

Radio rules and CSV schema · Credits