fullseye
tb_tcspc_irf_convolve — 2D typed op
- データ種:
counts→counts - 呼び出し:
fullseye.apply(img, "tb_tcspc_irf_convolve", 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)
図は合成の入力 128×128 で実際に走らせた出力。左が入力、右が出力。点群は上から見た散布(明るさ = z)、1-D 列は折れ線、体積は z 方向の最大値投影、動画は中央フレーム、複素画像は振幅、絵にならない返り値は値そのもの。
つまみ a を振る(0.1 / 0.5 / 0.9、もう一方は既定):
つまみ b を振る(0.1 / 0.5 / 0.9、もう一方は既定):
段階(前置きの op → この op。左から順):
別の画像でも(合成シーン / 写真 / 硬貨。上段が入力、下段がその出力。つまみは既定):
使い方
Blur an arrival-time histogram by the instrument response (timing jitter).
The temporal analogue of a PSF convolution: a detector's timing uncertainty
(SPAD jitter + TDC quantisation + laser pulse width) smears every arrival
time by the instrument response function, here a Gaussian of full width at
half maximum *irf_fwhm_ps*. The kernel is the **exact bin integral** of that
Gaussian (erf differences), normalised to sum 1, truncated at
``+-truncate*sigma`` and forced to odd length so the convolution is centred.
Ground truth: convolving a unit spike in the middle of a 256-bin window with
``irf_fwhm_ps = 500`` at ``bin_ps = 50`` leaves the centroid **exactly**
where it was (measured shift 0.0 ps — the kernel is symmetric) and gives a
profile whose measured FWHM is 501.22 ps. That 0.24% excess over 500 is the
*measurement*, not the kernel: :func:`tcspc_stats` finds the half-maximum
crossings by linear interpolation between bins, which slightly overestimates
the width of a Gaussian.
Total counts are preserved *except* at the window edges, where
``mode='same'`` discards the tail that falls outside — measured loss exactly
0 for that centred spike, but a genuine loss for a pulse within a few sigma
of either end.
Returns a float64 1-D histogram of the same length as *hist*.
**Raises** ``ValueError``: negative, non-finite or non-1-D *hist*, a
non-positive *bin_ps* / *irf_fwhm_ps* / *truncate*, an IRF sigma below
1e-3 bins (the kernel would be a delta and the op a no-op — say so instead
of pretending to blur), and a kernel that would be longer than the
:data:`MAX_BINS` cap.
Typed bridge of the photon op tcspc_irf_convolve into the 2-D evolution registry: the same implementation, called under the op(v, a, b) convention. a drives bin_ps (default 100) and b drives irf_fwhm_ps (default 200).
参考(サンプルデータ・文献)
- 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。
Studio で試す
下のプログラムは実際に走ることを確かめてある(図と同じ入力)。Studio のヘルプではこのブロックがボタンになり、その場で読み込んで実行できる。
img_to_counts 0.50 0.50
tb_tcspc_irf_convolve 0.50 0.50
実行できる例(この op を実際に呼ぶ検証済みサンプル)
次の例は元の台帳 op tcspc_irf_convolve を呼ぶもの。この橋渡し op は同じ実装を fn(v, a, b) 規約に合わせただけなので、挙動はそのまま当てはまる(呼び出し形だけ違う)。
- photon_timeresolved —
py -3.11 examples/photon_timeresolved.py
型が繋がる次の op(counts を入力に取れる)
identity · tb_spad_deadtime_apply · tb_spad_deadtime_correct · tb_tcspc_coates_correct · tb_tcspc_background_subtract · tb_dtof_depth · tb_countrate_to_counts · tb_counts_to_countrate
同カテゴリ(typed)
tb_points_to_voxel · tb_estimate_point_normals · tb_iss_keypoints · tb_project_points · tb_render_point_depth · tb_statistical_outlier_removal · tb_radius_outlier_removal · tb_voxel_grid_downsample
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.