decompose op• Data kinds: table → image2d
• Call: import motionmag; motionmag.complex_steerable_reconstruct(decomposition) -> 'np.ndarray' (or opsmotionmag.get("complex_steerable_reconstruct"))
Invert :func:complex_steerable_decompose -> `(H, W)` real image.
Exact: the analysis bank is a tight frame after the divisor correction, so
the round trip is the identity up to floating-point rounding (measured
`6.66e-16` maximum absolute error on a 64x64 random frame with the default
4 scales x 4 orientations). Editing a band before calling this — scaling its
phase, zeroing it — is how every other operator in this module works.
Refuses a dict whose bands do not match the declared shape or count, because
a partially edited decomposition would otherwise reconstruct something
plausible from the wrong frame size.
• motion_magnification family guide
• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).
• Operator provenance and references — the sources of the research/methods this op family came from.
• The canonical algorithm (author, year) and its uses are named in the family usage guide above.
• motion_magnification — py -3.11 examples/motion_magnification.py
image2d as input)decompose)*Provenance: motionmag.py — MOTIONMAG operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.