keypoints_from_image2d — REPRCONV `keypoint` op

• Data kinds: image2d → keypoints

• Call: import reprconv; reprconv.keypoints_from_image2d(image2d, threshold=0.0) (or opsreprconv.get("keypoints_from_image2d"))

Usage

A count/response image `(H, W) → image coordinates (N,2) = (u, v)`. The way back of the round trip.

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

`> threshold` の画素を 8 近傍で連結成分に分け、各成分の**強度重み付き

重心**を返す。重み付きにするのは、副画素の情報が残っている応答画像

(相関ピーク等)で往復誤差を量子化以下へ落とせるようにするため。

8 近傍の連結が損失の主犯である点に注意: 隣り合う画素に落ちた 2 点は

1 つの成分に融合し、重心が 2 点の中間へ動く。`selftest` は

「よく離れた点だけの量子化誤差」と「融合を含む全体」を別々に測る

(混ぜると量子化の理論値 0.2887 px と比較できなくなる)。

Args:

image2d: (H, W) の実画像。

threshold: この値を超えた画素だけを拾う。

Returns:

(N, 2) float64 の (u, v)。行順は `scipy.ndimage.label` のラベル順。

Raises:

ValueError: 2-D でない / 非有限 / 閾値を超える画素が 1 つも無い。

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.

• The canonical algorithm (author, year) and its uses are named in the family usage guide above.

Runnable examples (verified samples that actually call this op)

• representation_roundtrip — py -3.11 examples/representation_roundtrip.py

Ops the type connects to (they accept `keypoints` as input)

keypoints_uv_to_points · keypoints_to_image2d

Same category (`keypoint`)

keypoints_uv_to_points · points_zyx_to_keypoints_uv · keypoints_to_image2d · position_to_points · points_to_position

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

keypoints_from_image2d — REPRCONV keypoint op