tv_min.py¶

Source: LION/classical_algorithms/tv_min.py

Chambolle–Pock total-variation reconstruction.

LION.classical_algorithms.tv_min.cite(cite_format='MLA')[source]¶

Print the Chambolle-Pock primal-dual citation.

Parameters:

cite_format (str)

Return type:

None

LION.classical_algorithms.tv_min.tv_min(sino, op, lam, num_iterations=500, L=None, non_negativity=False, progress_bar=False, callbacks=[])[source]¶

Minimise a TV-regularised CT objective with Chambolle–Pock.

Parameters:

  • sino (torch.Tensor) – Batched sinograms in NCHW-like projection layout.

  • op (tomosipo.Operator or Geometry) – Forward projection operator or source geometry.

  • lam (float) – TV regularisation weight.

  • num_iterations (int, optional) – Chambolle–Pock iterations.

  • L (float, optional) – Precomputed operator norm used by the backend.

  • non_negativity (bool, optional) – Constrain reconstruction values to be non-negative.

  • progress_bar (bool, optional) – Display backend progress.

  • callbacks (list of callable, optional) – Iteration callbacks passed to ts_algorithms.tv_min2d.

Returns:

Batched TV reconstructions.

Return type:

torch.Tensor