typed op• 数据种类:lightfield → image
• 调用:fullseye.apply(img, "tb_lf_refocus", a=0.5, b=0.5)(2-D 的模型是一张图 + 两个标量旋钮 a,b∈[0,1])
*无图: 该算子以 lightfield 为输入。从图像开始的 Studio 程序无法到达该类型，请参见下方的可运行示例。*
位移叠加重聚焦：在指定 *slope* 处对焦的合成孔径图像。
> 以下的详细说明为原文 —— 摘要与标题已翻译。
Every view `(v, u) is shifted by (-s*(v - v_c), -s*(u - u_c))` — the
minus undoes the parallax of a point at slope `s` — and the shifted
views are averaged. Points at that slope add coherently and stay sharp;
everything else is smeared by an amount proportional to its slope
difference times the angular baseline. `slope=0` is the plane the array
was already focused on and returns the plain average of the views.
Ground truth it reproduces exactly (pinned in `tests/test_lightfield.py`):
a single-layer field synthesised at slope `s0 and refocused at s0`
with `edge="wrap" and an integer s0` returns the original texture to
5.6e-16; sweeping the slope, the variance of the result peaks at `s0`
(measured exactly on the sweep grid in all 18 texture/slope combinations
listed in the module docstring), and refocusing at `-s0` does *not* —
which is the check that catches a flipped shift sign.
Returns a `(H, W)` 2-D image.
Raises `ValueError`: *lf* not a valid light field, a non-finite or
over-large *slope* (:data:MAX_ABS_SLOPE), unknown *interp* / *edge*.
Typed bridge of the lightfield op `lf_refocus into the 2-D evolution registry: the same implementation, called under the op(v, a, b) convention. This op has no tunable parameter; a and b` are unused.
• 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。
• (暂无)
image 作为输入)identity · gaussian · mean_box · bilateral · unsharp · median · min_filter · max_filter
typed)tb_points_to_voxel · tb_estimate_point_normals · tb_iss_keypoints · tb_angle_3points · tb_project_points · tb_render_point_depth · tb_statistical_outlier_removal · tb_radius_outlier_removal
*Provenance: ops.py — 2D 算子登记表。本条目由 tools/opdocs.py md 自动生成(请勿手工编辑)。*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.