lf_from_mla — LIGHTFIELD `decode` op

• 数据种类:image2d → lightfield

• 调用:import lightfield; lightfield.lf_from_mla(raw, angular=(5, 5), *, offset=(0, 0), crop=False)(或 opslightfield.get("lf_from_mla"))

把微透镜阵列的原始帧解码成 `(V, U, H, W)` 光场。

> 以下的详细说明为原文 —— 摘要与标题已翻译。

The plenoptic sensor stores, for microlens `(t, s), a V x U` block of

pixels — one per direction through the main lens' exit pupil. So raw pixel

`(offset_y + t*V + v, offset_x + s*U + u) is exactly L[v, u, t, s]`:

the decode is a pure re-sort, no interpolation, no data invented.

:func:lf_to_mla inverts it bit-exactly.

*offset* is the integer position of the first whole microlens; a real MLA is

never aligned to pixel `(0, 0)`. Sub-pixel MLA calibration is out of scope

(see the module docstring) — this operator works on an already-rectified

frame.

By default a raw frame whose usable size is not a whole multiple of the

microlens pitch is a `ValueError`, because the alternative — quietly

dropping the last partial row of microlenses — shifts every subsequent

microlens centre and produces a light field that looks right and is wrong.

Pass `crop=True` to opt in to that crop explicitly.

Raises `ValueError`: *raw* not 2-D or non-finite, *angular* outside

`[1, MAX_ANGULAR]`, a negative *offset*, an *offset* that leaves fewer

than one whole microlens, a size that is not a multiple of the pitch (unless

`crop=True), a decoded sub-aperture image over :data:MAX_SPATIAL` on a

side, and a decoded size over :data:MAX_LF_ELEMENTS.

• 示例数据目录(下载 URL / 许可证) —— 2-D 用 skimage.data(BSD/公有领域)加合成图,3-D 给出真实数据源(Stanford/PDS 等)的下载 URL。

• 算子来历与参考文献 —— 该算子族所依据的研究/方法出处。

• 算法的正典(作者・年份)与用途见上面的族使用指南。

• lightfield_depth — py -3.11 examples/lightfield_depth.py

*Provenance: lightfield.py — LIGHTFIELD 算子登记表。本条目由 tools/opdocs.py md 自动生成(请勿手工编辑)。*

lf_from_mla — LIGHTFIELD decode op