fuse — 3D tsdf_fusion op

Data kinds: depthsdf

Call: import fullseye as fs; fs.ledger.fuse(depths: 'Sequence[np.ndarray]', Ks: 'Sequence', Rs: 'Sequence', ts: 'Sequence', bounds: 'Bounds', res: 'int', trunc: 'float') -> 'Tuple[np.ndarray, np.ndarray]' (to call the implementation directly, import tsdf_fusion; tsdf_fusion.fuse(depths: 'Sequence[np.ndarray]', Ks: 'Sequence', Rs: 'Sequence', ts: 'Sequence', bounds: 'Bounds', res: 'int', trunc: 'float') -> 'Tuple[np.ndarray, np.ndarray]'; from the registry, ops3d.get("fuse"))

Usage

Fuse a depth series into one TSDF volume with new_volume + integrate. Returns (tsdf, weight).

> The detailed description below is the original text — the summary and the headings are translated.

各フレームの (K,R,t) は world→camera。多視点で同じ表面を観測すると重みが積算され、

単フレームでは見えない(自己遮蔽の)面が別視点で埋まる。

• `depths`: (H,W) 深度画像のリスト(カメラ Z 深度。0 以下・非有限の画素は未観測扱い)。

サイズはフレームごとに異なってよい。

• `Ks / Rs / ts: 各フレームの (3,3) 内部行列・(3,3) 回転・(3,) 並進(X_cam = R X + t`)。

• `bounds: ((xmin,xmax),(ymin,ymax),(zmin,zmax))` の world 領域(各軸 max > min)。

• `res: 一辺の voxel 数(正の int)。volume は res×res×res、軸順は (x,y,z) の indexing='ij'`。

• `trunc`: 切り詰め距離(world 単位、正の有限値)。voxel サイズの数倍が目安で、小さすぎると

表面の両側に観測が乗らず、大きすぎると薄い構造の表裏が混ざる。

返り値: `tsdf (res,res,res) float32(未観測 +1.0、表面手前が正・奥が負、[-1,1])と weight`

(res,res,res) float32(観測回数)。表面点は `extract_surface_points(tsdf, weight, bounds, res)`

で取り出す。fail-closed: フレーム 0 枚、リスト長の不一致、退化 bounds、非正の `res` /

`truncValueError。統合の詳細(遮蔽領域を更新しない規則)は integrate` を参照。

Background guides (the physics and conventions behind this op)

depth_sensors — 深度センサの知識 — 測距原理・実機の値・欠測の出方

References (sample data, literature)

• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).

• Operator provenance and references — the sources of the research/methods this op family came from.

Runnable examples (verified samples that actually call this op)

tsdf_fusion_demopy -3.11 examples_3d/tsdf_fusion_demo.py

Ops the type connects to (they accept sdf as input)

sdf_to_occupancy · fuse_to_voxel · integrate · extract_surface_points · query_distance · sdf_union · sdf_intersect · sdf_subtract

Same category (tsdf_fusion)

integrate · extract_surface_points


*Provenance: tsdf_fusion.py — 3D operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*

© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.