geometry op• Datenarten: sinogram → sinogram
• Aufruf: import tomography; tomography.sinogram_center_shift(sinogram, shift_px=None, angles_deg=None) (oder opstomography.get("sinogram_center_shift"))
Zentriert ein Sinogramm neu auf seine Rotationsachse.
> Die ausführliche Beschreibung unten ist der Originaltext — Zusammenfassung und Überschriften sind übersetzt.
Shifts every projection by `-shift_px` along the detector axis with linear
interpolation. With `shift_px=None` the shift is measured first by
:func:sinogram_center_of_rotation, which makes this the one-call fix.
Round-trip on the Shepp-Logan phantom, shifting by *d* and back:
d max |error| relative to the peak line integral
1.00 px 0.0e+00 0.0e+00 (an integer shift is exact)
0.50 px 1.4e-01 1.2e-01
0.25 px 1.1e-01 9.2e-02
A *fractional* shift is not a small operation and this is the operator's
honest limitation: 12 % of the peak, on a phantom with sharp edges, from one
round trip. Interpolation is a low-pass filter and the sinogram of an edge is
not band-limited, so there is nothing to recover on the way back. Using a
Fourier shift instead would trade this visible blur for invisible ringing at
the detector edges, which is worse in the way that matters here. The
consequence is in :func:sinogram_center_of_rotation's table: an integer
centre error is fully repairable, a half-pixel one is not.
:param sinogram: `(n_angles, n_detectors)`.
:param shift_px: axis offset in detector bins; `None` -> measure it.
:param angles_deg: view angles, used only when *shift_px* is `None`.
:returns: `(n_angles, n_detectors)` float64 sinogram.
:raises ValueError: if `|shift_px|` is at or past half the detector width —
past that the object has been shifted out of the field of view and what
comes back is edge padding, which reconstructs as a plausible, empty slice.
• 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.
• ct_reconstruction — py -3.11 examples/ct_reconstruction.py
sinogram als Eingabe)backproject_sinogram · filtered_backprojection · sart_reconstruct · beam_hardening_apply · beam_hardening_correct · ring_artifact_apply · ring_artifact_remove · metal_trace_interpolate
geometry)*Provenance: tomography.py — TOMOGRAPHY 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.