"""
DeepNIR vs 传统 NIR 传感器性能对比

DeepNIR 核心改进：
1. 像素结构优化 → 量子效率提升 60%
2. 嵌入式处理 → 减少 ISP 负担
3. MTF 锐度提升 → 更清晰图像
4. 紧凑封装 → 更小摄像头模组
"""

import numpy as np

# 传感器参数对比
sensor_specs = {
    'VD56G DeepNIR': {
        'quantum_efficiency': 0.65,   # 65% QE @ 940nm
        'mtf': 0.85,                  # MTF
        'read_noise_e': 1.5,          # 读出噪声 (e-)
        'dark_current': 0.5,          # 暗电流 (e-/s)
        'pixel_size_um': 2.8,         # 像素尺寸
        'embedded_isp': True,         # 嵌入式 ISP
        'package': 'CSP',             # 封装
    },
    '传统 NIR 传感器': {
        'quantum_efficiency': 0.40,   # 40% QE
        'mtf': 0.63,
        'read_noise_e': 2.5,
        'dark_current': 1.2,
        'pixel_size_um': 3.0,
        'embedded_isp': False,
        'package': 'CSP',
    },
}

# 940nm 红外场景下 SNR 对比
def calculate_snr(qe, read_noise, dark_current, 
                   light_power, exposure_ms):
    """
    计算传感器信噪比
    
    SNR = (N_signal) / sqrt(N_signal + N_read^2 + N_dark)
    """
    # 光子数
    photon_rate = light_power * 1e-3 * 0.65  # 简化计算
    n_photons = photon_rate * exposure_ms * 1e-3
    
    # 信号电子数
    n_signal = n_photons * qe
    
    # 暗电流电子数
    n_dark = dark_current * exposure_ms * 1e-3
    
    # 总噪声
    noise = np.sqrt(n_signal + read_noise**2 + n_dark)
    
    snr = n_signal / (noise + 1e-10)
    return 20 * np.log10(snr)  # dB

# 不同光照条件下对比
light_conditions = {
    '白天 (高亮度)': 1000,
    '黄昏 (中亮度)': 100,
    '夜间红外补光 (低亮度)': 10,
    '极暗 (仅红外)': 1,
}

print("=== DeepNIR vs 传统 NIR SNR 对比 (940nm) ===")
print(f"{'场景':<25} {'传统NIR(dB)':>12} {'DeepNIR(dB)':>12} {'提升(dB)':>10}")
print("-" * 65)

for condition, power in light_conditions.items():
    snr_trad = calculate_snr(
        sensor_specs['传统 NIR 传感器']['quantum_efficiency'],
        sensor_specs['传统 NIR 传感器']['read_noise_e'],
        sensor_specs['传统 NIR 传感器']['dark_current'],
        power, 33  # 33ms 曝光
    )
    snr_deep = calculate_snr(
        sensor_specs['VD56G DeepNIR']['quantum_efficiency'],
        sensor_specs['VD56G DeepNIR']['read_noise_e'],
        sensor_specs['VD56G DeepNIR']['dark_current'],
        power, 33
    )
    improvement = snr_deep - snr_trad
    print(f"{condition:<25} {snr_trad:>12.1f} "
          f"{snr_deep:>12.1f} {improvement:>+10.1f}")

print("\n=== DeepNIR 关键优势 ===")
print("1. 量子效率 +60% → 暗光下信号更强")
print("2. MTF 锐度 +35% → 图像更清晰")
print("3. 读出噪声 -40% → SNR 更高")
print("4. 暗电流 -58% → 长曝光噪声更小")
print("5. 嵌入式 ISP → 简化系统设计")

"""
DeepNIR 对 PERCLOS 测量精度的影响分析

PERCLOS = Percentage of Eye Closure
在低光条件下，传统 NIR 传感器噪声大，
导致眼睑检测精度下降，PERCLOS 误差增大
"""

# 模拟 PERCLOS 在不同传感器下的误差
conditions = {
    '白天': {'traditional': 2.1, 'deepnir': 1.5},  # 误差 %
    '黄昏': {'traditional': 5.3, 'deepnir': 2.0},
    '夜间': {'traditional': 12.8, 'deepnir': 3.2},
    '极暗': {'traditional': 25.0, 'deepnir': 5.8},
}

print("=== PERCLOS 测量误差对比 ===")
print(f"{'场景':<10} {'传统NIR(%)':>12} {'DeepNIR(%)':>12} {'改善':>10}")
print("-" * 50)
for cond, data in conditions.items():
    improvement = data['traditional'] - data['deepnir']
    print(f"{cond:<10} {data['traditional']:>12.1f} "
          f"{data['deepnir']:>12.1f} {improvement:>+10.1f}%p")