1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
| import numpy as np
from typing import Tuple
class DMSImageQualityTest:
"""
基于 IEEE-2020:2024 的 DMS 摄像头图像质量测试
Vega IR 支持的关键测试项
"""
def __init__(self, vega_ir_illuminance: float = 1.0):
"""
Args:
vega_ir_illuminance: Vega IR 辐照度 (W/m²)
"""
self.illuminance = vega_ir_illuminance
def measure_cta(
self,
image: np.ndarray,
target_roi: Tuple[int, int, int, int] = None
) -> float:
"""
Contrast Transfer Accuracy (CTA)
IEEE-2020:2024 定义
衡量摄像头在 NIR 下对比度还原能力
CTA = (L_max - L_min) / (L_max + L_min)
良好: CTA > 0.45
可接受: CTA > 0.30
不合格: CTA < 0.30
Args:
image: 测试图卡图像
target_roi: (x, y, w, h) 感兴趣区域
Returns:
cta: 对比度传递精度值
"""
if target_roi:
x, y, w, h = target_roi
roi = image[y:y+h, x:x+w]
else:
roi = image
if len(roi.shape) == 3:
roi = np.mean(roi, axis=2)
h, w = roi.shape
quarter_h = h // 4
quarter_w = w // 4
l_max = np.percentile(roi[quarter_h:3*quarter_h, quarter_w:3*quarter_w], 99)
l_min = np.percentile(roi[quarter_h:3*quarter_h, quarter_w:3*quarter_w], 1)
if l_max + l_min == 0:
return 0.0
cta = (l_max - l_min) / (l_max + l_min)
return cta
def measure_csnr(
self,
image: np.ndarray,
dark_frames: np.ndarray = None,
target_roi: Tuple[int, int, int, int] = None
) -> float:
"""
Contrast Signal-to-Noise Ratio (CSNR)
IEEE-2020:2024 定义
CSNR = 20 * log10(Contrast / Noise_Std)
良好: CSNR > 35 dB
可接受: CSNR > 25 dB
不合格: CSNR < 25 dB
Args:
image: 测试图卡图像
dark_frames: 暗帧序列（用于固定模式噪声）
target_roi: ROI
Returns:
csnr_db: CSNR (dB)
"""
if target_roi:
x, y, w, h = target_roi
roi = image[y:y+h, x:x+w]
else:
roi = image
if len(roi.shape) == 3:
roi = np.mean(roi, axis=2)
quarter_h = roi.shape[0] // 4
quarter_w = roi.shape[1] // 4
center = roi[quarter_h:3*quarter_h, quarter_w:3*quarter_w]
contrast = np.mean(center) - np.std(center)
noise_std = np.std(center)
if noise_std == 0:
return float('inf')
csnr = 20 * np.log10(max(contrast, 1) / noise_std)
return csnr
def measure_sfr_nir(
self,
image: np.ndarray,
edge_roi: Tuple[int, int, int, int]
) -> Tuple[np.ndarray, float]:
"""
NIR 下的 MTF/SFR 测量
使用斜边法（ISO 12233 适配 NIR）
Args:
image: 测试图卡图像（含斜边）
edge_roi: 斜边区域
Returns:
sfr_curve: SFR 曲线
sfr_at_nyquist: Nyquist 频率处的 SFR 值
"""
x, y, w, h = edge_roi
edge_region = image[y:y+h, x:x+w]
if len(edge_region.shape) == 3:
edge_region = np.mean(edge_region, axis=2)
row_means = np.mean(edge_region, axis=1)
edge_position = np.argmax(np.abs(np.diff(row_means)))
esf = row_means[max(0, edge_position-20):edge_position+20]
lsf = np.diff(esf)
sfr = np.abs(np.fft.fft(lsf))
sfr = sfr / sfr[0]
nyquist_idx = len(sfr) // 2
sfr_nyquist = sfr[nyquist_idx]
return sfr, sfr_nyquist
if __name__ == "__main__":
tester = DMSImageQualityTest(vega_ir_illuminance=1.0)
np.random.seed(42)
test_chart = np.random.randint(0, 255, (960, 1280), dtype=np.uint8)
cta = tester.measure_cta(test_chart, (200, 200, 400, 400))
print(f"CTA: {cta:.3f} ({'✅ 良好' if cta > 0.45 else '⚠️ 可接受' if cta > 0.30 else '❌ 不合格'})")
csnr = tester.measure_csnr(test_chart, target_roi=(200, 200, 400, 400))
print(f"CSNR: {csnr:.1f} dB ({'✅ 良好' if csnr > 35 else '⚠️ 可接受' if csnr > 25 else '❌ 不合格'})")
sfr, sfr_nyq = tester.measure_sfr_nir(test_chart, (100, 100, 200, 800))
print(f"SFR@Nyquist: {sfr_nyq:.3f} ({'✅ 良好' if sfr_nyq > 0.3 else '⚠️ 需检查'})")
|
