texture op• Data kinds: image → image
• Call: fullseye.apply(img, "gabor", a=0.5, b=0.5) (the 2-D model is one image plus two scalar knobs a,b∈[0,1])
• HALCON equivalent: gen_gabor (the HALCON reference is a useful guide to its meaning and parameters)
*The figure is the real output on a synthetic 128×128 input. Left: input, right: output (a non-image return value is shown as the value itself).*
Gabor energy |v * g| — an oriented band-pass texture response.
> The detailed description below is the original text — the summary and the headings are translated.
• `a — **向き** θ = π·a` [rad]。カーネルの余弦は回転後の x 方向に走るので、
`a=0` (θ=0) は 縦縞(列方向に明暗が変わる模様)に最も強く応答し、
`a=0.5 (θ=90°) は **横縞**に応答する。a=1` は θ=180° で a=0 と同じ向き。
• `b — 空間周波数 0.1 + 0.3·b` [cycles/px]。
★正規化(2026-09-02 の修正): カーネルの L1 ノルムで割る固定スケール。
以前は `_norm` = その画像での最大絶対値で割っていたため、**応答の大小そのもの
が消えていた**。実測(96×96 の横縞、周波数 0.25): 生の畳み込みの平均振幅は
θ=0° が 0.0165、θ=90° が 0.9077 で 54.9 倍の差があるのに、`_norm` を通すと
平均は 0.3554 対 0.4790 = 1.35 倍まで潰れていた —— 向きを見分けるための
特徴量なのに識別力が消えていた(向きごとに別の除数で割っていたのだから当然)。
`|v| <= 1 なら |v * g| <= sum|g|` なので L1 で割れば値域 [0,1] を保ったまま
op を跨いで比較できる絶対スケールになる。
• gallery2d_texture_freq family guide
• 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.
The program below has been verified to run (same input as the figure). In Studio's help this block becomes buttons that load and run it on the spot.
gabor 0.50 0.50
▸ Load this pipeline · Load & run
• gallery2d_texture_freq — py -3.11 examples/gallery2d_texture_freq.py
image as input)identity · gaussian · mean_box · bilateral · unsharp · median · min_filter · max_filter
texture)std_filter · sk_frangi · sk_meijering · sk_hessian · sk_gabor · sk_lbp · sk_entropy · sk_shape_index
*Provenance: ops.py — 2D 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.