lattice_moire_peak — FLYVISION alias op

• 数据种类:signal × table → table

• 调用: import fullseye as fs; fs.ledger.lattice_moire_peak(signal, lattice, fmax_cycles_per_deg=0.0, n_freq=48, window='hann')(要直接调用实现,import flyvision; flyvision.lattice_moire_peak(signal, lattice, fmax_cycles_per_deg=0.0, n_freq=48, window='hann');从台账取用则 opsflyvision.get("lattice_moire_peak"))

用法

采样信号中最强低频成分的尖锐程度——晶格对无法分辨的纹理所产生的莫尔条纹。

> 以下的详细说明为原文 —— 摘要与标题已翻译。

Sampling a texture above the lattice's Nyquist limit does not lose it; it folds it to a lower frequency. On a regular lattice the fold is a single sharp peak — a false stripe the next stage reads as a real one — and on an irregular arrangement the same energy spreads into a ring. This op measures that difference as a number: the non-uniform discrete Fourier transform of the ommatidial signal (mean removed, radially windowed) at the gnomonic positions of the ommatidia, over a grid of n_freq x n_freq frequencies up to *fmax_cycles_per_deg* (0 = the lattice's own hexagonal Nyquist 1/(sqrt 3 dphi), the band the eye can represent), and the ratio of the highest amplitude on that grid to the median amplitude.

signal: (n,) ommatidial values (fly_view_plane_texture).

lattice: the eye they were sampled on — any geometry; only directions are used. fmax_cycles_per_deg: the half-width of the frequency grid, cycles/degree.

n_freq: grid side, 8 <= n_freq <= MAX_MOIRE_FREQ.

window: one of MOIRE_WINDOWS — a radial Hann taper over the field

(default: the field edge is a step and would leak everywhere).

Returns a dict {"peak_ratio", "peak_cycles_per_deg", "peak_amplitude", "background_median", "nyquist_cycles_per_deg", "n_freq"}.

Ground truth (pinned in the tests): a sinusoid just below Nyquist sampled on the hexagonal eye gives a peak ratio above 30; the same sinusoid on a Vogel eye of the same count and mean spacing gives a peak ratio at least 2.5x smaller — the energy that was one peak is now a ring.

Raises ValueError: a *signal* that is not (n,) for this lattice (an empty *signal* is refused), non-finite entries, a constant signal (there is no spectrum to rank — its ratio would be 0/0), a negative *fmax*, an *n_freq* outside its range, an unknown *window*, and a malformed *lattice*.

详细使用指南

• fly_vision 族使用指南

参考(示例数据、文献)

• 示例数据目录(下载 URL / 许可证) —— 2-D 用 skimage.data(BSD/公有领域)加合成图,3-D 给出真实数据源(Stanford/PDS 等)的下载 URL。

• 算子来历与参考文献 —— 该算子族所依据的研究/方法出处。

• 算法的正典(作者、年份)与用途见上面的族使用指南。

可运行的示例(实际调用该算子并已验证的样例)

• poc_moire_screen — py -3.11 examples/poc_moire_screen.py

• poc_print_registration — py -3.11 examples/poc_print_registration.py

• poc_zebra_stripes_fly_eye — py -3.11 examples/poc_zebra_stripes_fly_eye.py

类型可衔接的下一个算子(可接受 table 作为输入)

fly_t4t5_field · fly_flow_from_directions · fly_matched_filter · fly_egomotion_from_flow · fly_eye_merge · fly_view_plane_texture · fly_hex_resample · fly_expansion_readout

同类别(alias)

sampling_alias_distance


*Provenance: flyvision.py — FLYVISION 算子登记表。本条目由 tools/opdocs.py md 自动生成(请勿手工编辑)。*

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