关于「空间频率:细条纹先走,剪影最后离场」,《服饰图案远距辨识与再认》给出的定量结论是什么?

结论:第二章把图案周期折成每度周数,4 毫米排线五米外只剩 0.092 的调制,剪影却以平均亮度留下。

数据口径:书中参数为示意,全部计算由书中程序生成。 本书不是商标或著作权法律意见,示意图案全部为作者自拟几何形,不点名评价任何真实品牌,不写任何潜意识影响宣称,街头测试只记录匿名对错,不采集人脸与个人资料。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 0.092

d(m) sigma(mm)4mm: numanal20mm: numanalmean
10.290.9981.0001.0001.0000.550
30.870.4910.4981.0001.0000.550
51.450.0920.0940.9991.0000.550
102.910.0000.0000.8280.8390.550
154.360.0000.0000.4970.4980.550
查看完整验算程序
# Chapter 2. p23: a distance blur erases hatching before the silhouette.
# Model: a 1-D luminance profile across the author's emblem on a dark
# tee: light silhouette from 20 to 280 mm with 4 mm hatching inside the
# left half and 20 mm bars inside the right half. A Gaussian blur of
# angular sigma 1 arcmin (illustrative: optics, focus, motion) becomes
# sigma = d tan(1') in mm. Residual modulation of a period P is
# exp(-2 pi^2 sigma^2 / P^2) analytically; the numeric convolution on a
# 0.25 mm grid checks it. Mean: the hatched half keeps its average
# luminance against the 0.05 shirt, so the silhouette survives.
import math

DX = 0.25
X = [(i + 0.5) * DX for i in range(int(300 / DX))]
ARCMIN = math.pi / (180.0 * 60.0)


def profile(x):
    if x < 20.0 or x > 280.0:
        return 0.05
    if x < 150.0:
        return 0.80 if math.fmod(x, 4.0) < 2.0 else 0.30
    return 0.80 if math.fmod(x, 20.0) < 10.0 else 0.30


SIG = [profile(x) for x in X]


def blur(sig, sigma):
    r = int(4.0 * sigma / DX) + 1
    w = [math.exp(-0.5 * ((k * DX) / sigma) ** 2) for k in range(-r, r + 1)]
    tot = math.fsum(w)
    out = []
    n = len(sig)
    for i in range(n):
        terms = [w[k + r] * sig[min(max(i + k, 0), n - 1)]
                 for k in range(-r, r + 1)]
        out.append(math.fsum(terms) / tot)
    return out


def swing(out, a, b):
    seg = out[int(a / DX):int(b / DX)]
    return max(seg) - min(seg)


if __name__ == "__main__":
    base4, base20 = 0.5, 0.5
    print(" d(m) sigma(mm)  4mm: num  anal   20mm: num  anal  mean")
    for d in (1.0, 3.0, 5.0, 10.0, 15.0):
        s = d * math.tan(ARCMIN) * 1000.0
        out = blur(SIG, s)
        n4 = swing(out, 60.0, 100.0) / base4
        n20 = swing(out, 190.0, 250.0) / base20
        # square wave fundamental keeps 4/pi of the swing before blur
        a4 = min(1.0, 4.0 / math.pi * math.exp(
            -2.0 * math.pi ** 2 * s ** 2 / 16.0))
        a20 = min(1.0, 4.0 / math.pi * math.exp(
            -2.0 * math.pi ** 2 * s ** 2 / 400.0))
        edge = math.fsum(out[int(60.0 / DX):int(100.0 / DX)]) / (40.0 / DX)
        print(" {:>4.0f} {:>8.2f}   {:>6.3f} {:>6.3f}   {:>6.3f} {:>6.3f}"
              "  {:.3f}".format(d, s, n4, a4, n20, a20, edge))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《服饰图腾完形视知觉捕获、远距高对比辨识与街头视觉模因驻留母典》第 2 章「空间频率:细条纹先走,剪影最后离场」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-390 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026390,
  author    = {Fan, Yafu},
  title     = {Apparel Graphic Gestalt Capture, Long-Range Contrast Legibility and Streetwear Visual Memetic Retention},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论