process op• Datenarten: beatcube → image2d
• Aufruf: import rangedoppler; rangedoppler.range_doppler_map(cube, combine='incoherent', antenna=None, normalize=False) (oder opsrangedoppler.get("range_doppler_map"))
Die 2-D-FFT eines Beat-Würfels -> eine `(n_doppler, n_range)`-Betragskarte.
> Die ausführliche Beschreibung unten ist der Originaltext — Zusammenfassung und Überschriften sind übersetzt.
Fast time transforms to range (last axis, not shifted: bin `j` is
`j * c*f_s/(2*S*N_s)` metres, and a physical range is always positive so
the whole `[0, f_s)` band is used). Slow time transforms to velocity
(middle axis, `fftshift`ed so the map is centred on zero velocity: bin
`i is (i - N_c//2) * lambda/(2*N_c*T_c)` metres per second, positive =
receding).
The antenna axis is collapsed by *combine*: `"incoherent"` (default) is the
mean of the magnitudes, which is angle independent and therefore the
right default for detection; `"coherent"` is the magnitude of the mean,
i.e. a beam pointed at boresight, which attenuates an off-boresight target on
purpose. `antenna=k uses element k` alone. For a single-element cube
all three agree exactly.
`normalize=True divides by N_c * N_s`, so a bin-centred target of
amplitude `a peaks at exactly a` (measured: 1.0 for a unit target,
absolute error 0.0). The default `False` keeps the raw FFT magnitude.
No window is applied — compose :func:fmcw_window_apply first if you want
one. The output is a plain 2-D float64 array, so every 2-D operator in
Fullseye (threshold, morphology, labelling, blob measurement — the pieces a
CFAR detector is made of) applies to it directly.
Raises `ValueError`: a real-valued cube (it would put a mirror ghost of
every target at a fabricated range), fewer than 2 chirps or 2 samples, an
out-of-range *antenna* index, an unknown *combine*, a cube over the element
cap, an FFT that overflows to NaN, or NaN/Inf on the way in.
• Leitfaden zur Familie fmcw_range_doppler
• Katalog der Beispieldaten (Download-URLs / Lizenzen) — 2-D nutzt skimage.data (BSD/Public Domain) plus synthetische Bilder, 3-D nennt Download-URLs echter Datenquellen (Stanford, PDS, …).
• Herkunft und Literatur der Operatoren — die Quellen der Forschung/Verfahren, auf denen diese Operatorfamilie beruht.
• Der kanonische Algorithmus (Autor, Jahr) und seine Anwendungen stehen im Familienleitfaden oben.
• fmcw_range_doppler — py -3.11 examples/fmcw_range_doppler.py
image2d als Eingabe)process)fmcw_window_apply · range_doppler_peaks · fmcw_range_profile
*Provenance: rangedoppler.py — RANGEDOPPLER Operator-Registry. Diese Notiz wird von tools/opdocs.py md erzeugt (nicht von Hand bearbeiten).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.