关于「记忆驻留:三次偶遇不等于三次曝光」,《服饰图案远距辨识与再认》给出的定量结论是什么?
结论:第六章把「深印脑海」改写成再认概率:同样三次,分散曝光一个月后的保持是集中曝光的 2.38 倍;街头偶遇里,高对比图案有 0.311 的路人真正看见三次,低对比图案只有 0.027。
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。
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: 0.027, 0.311
| design meet>=3 caught>=3 R day 21: P10 P50 P90 mean | ||||
| D1 | 0.788 | 0.311 | 0.000 0.383 0.645 | 0.374 |
| D2 | 0.788 | 0.027 | 0.000 0.000 0.391 | 0.152 |
| expected meetings in 14 days: 4.2 | ||||
查看完整验算程序
# Chapter 6. p63: three street encounters are not three exposures.
# Model (illustrative): over 14 days a passer-by meets the emblem in a
# Poisson stream of LAM encounters a day; the distance of each meeting
# is uniform on 3 to 15 m; each meeting is captured with the logistic
# chance of p43 for design D1 (CR 9.13) or D2 (CR 1.36). Captured
# meetings feed the memory model of p62 (each first capture sets tau
# = 1 day; later captures multiply tau by 1 + G (1 - R)). Recognition
# is read on day 21. Seeded Monte Carlo over 20,000 passers-by.
import math
import random
LAM, DAYS, TEST, PEOPLE = 0.3, 14.0, 21.0, 20000
B0, B1, B2, B3 = -4.0, 1.0, 2.0, 1.5
PSI, G = 0.5, 3.0
def p_capture(size_cm, d_m, cr, k):
th = math.degrees(2.0 * math.atan((size_cm / 100.0) / (2.0 * d_m)))
z = B0 + B1 * math.log(th) + B2 * math.log10(cr) + B3 * k
return 1.0 / (1.0 + math.exp(-z))
def person(rng, cr):
t, meets, caught = 0.0, 0, []
while True:
t += rng.expovariate(LAM)
if t > DAYS:
break
meets += 1
d = rng.uniform(3.0, 15.0)
if rng.random() < p_capture(26.0, d, cr, 0.9):
caught.append(t)
if not caught:
return meets, 0, 0.0
tau, last = 1.0, caught[0]
for t in caught[1:]:
r = (1.0 + (t - last) / tau) ** (-PSI)
tau *= 1.0 + G * (1.0 - r)
last = t
return meets, len(caught), (1.0 + (TEST - last) / tau) ** (-PSI)
def q(xs, f):
return xs[min(len(xs) - 1, int(f * len(xs)))]
if __name__ == "__main__":
print("design meet>=3 caught>=3 R day 21: P10 P50 P90 mean")
for name, cr, rng in (("D1", 9.13, random.Random(39063)),
("D2", 1.36, random.Random(39064))):
m3, c3, rs = 0, 0, []
for _ in range(PEOPLE):
m, c, r = person(rng, cr)
m3 += m >= 3
c3 += c >= 3
rs.append(r)
rs.sort()
print("{:<6} {:>8.3f} {:>10.3f} {:>16.3f} {:>5.3f} {:>5.3f}"
" {:>6.3f}".format(name, m3 / PEOPLE, c3 / PEOPLE,
q(rs, 0.1), q(rs, 0.5), q(rs, 0.9),
math.fsum(rs) / PEOPLE))
print("expected meetings in 14 days: {:.1f}".format(LAM * DAYS))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 6 章「记忆驻留:三次偶遇不等于三次曝光」(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}
}