deformation op• データ種: image → image
• 呼び出し: fullseye.apply(img, "deform_tps", a=0.5, b=0.5) (2-D は 1 画像 + 2 スカラつまみ a,b∈[0,1] のモデル)
Thin-plate-spline warp over a 5x5 control grid (Bookstein, TPAMI 1989).
The 3x3 interior control points are displaced by the deterministic smooth
field `d = amp * [sin(2 pi f gx), cos(2 pi f gy)]` (gy, gx = the control
point's normalised coordinates), the 16 border control points are pinned so
the frame stays anchored. The backward map is the TPS interpolant fitted
from the *displaced* control points back to the original ones -- the unique
minimum-bending-energy interpolant, i.e. the surface a thin metal plate
would take -- and it is evaluated at every destination pixel before bilinear
resampling. `a sets the amplitude amp = 0.15*a*min(H,W), b` the
spatial frequency `f = 0.5 + 1.5b. a = 0` leaves the control points
where they are, so the solved map is the identity (up to sub-pixel resampling error).
• サンプルデータ カタログ(DL URL / ライセンス) — 2-D は skimage.data(BSD/public)+ 合成、3-D は実データ源(Stanford/PDS 等)の DL URL。
• 演算子の来歴・参考文献 — この op 族の元になった研究/手法の出典。
• アルゴリズムの正典(著者・年)と用途は上記ファミリ使い方ガイドに記載。
• gallery2d_geometry — py -3.11 examples/gallery2d_geometry.py
image を入力に取れる)identity · gaussian · mean_box · bilateral · unsharp · median · min_filter · max_filter
deformation)*Provenance: ops.py — 2D operator registry. この per-op ノートは tools/opdocs.py md が自動生成(手編集しない)。*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.