📡 多光谱成像

🌍 课程导言

人眼只能看到可见光(400-700nm)，但作物在近红外、红边等波段的反射携带了丰富的生理信息。多光谱成像让机器人能够"看见"叶绿素含量、水分胁迫、氮素状况等肉眼不可见的指标，实现早期预警。

📐 多光谱成像基础

💻 Python仿真

#!/usr/bin/env python3
"""多光谱成像仿真 - 植被指数计算与胁迫检测"""
import math, random
from collections import defaultdict

class MultispectralImage:
    """多光谱图像仿真"""
    def __init__(self, width=80, height=60, seed=42):
        self.w, self.h = width, height
        self.rng = random.Random(seed)
        self.bands = {}
        self._generate()
    
    def _generate(self):
        for name in ['blue','green','red','rededge','nir','swir']:
            self.bands[name] = [[0.0]*self.w for _ in range(self.h)]
        
        # 健康作物典型反射率
        for r in range(self.h):
            for c in range(self.w):
                health = 0.7 + 0.2 * math.sin(r*0.1) * math.cos(c*0.08)
                # 模拟胁迫区域
                if 20 < r < 35 and 40 < c < 60:
                    health *= 0.5  # 水分胁迫区
                if 45 < r < 55 and 10 < c < 25:
                    health *= 0.3  # 严重胁迫
                
                health = max(0.1, min(1.0, health + self.rng.gauss(0, 0.03)))
                
                self.bands['blue'][r][c] = max(0, 0.05 * (1-health) + self.rng.gauss(0, 0.005))
                self.bands['green'][r][c] = max(0, 0.10 * health + self.rng.gauss(0, 0.005))
                self.bands['red'][r][c] = max(0, 0.05 * (1-health) + self.rng.gauss(0, 0.005))
                self.bands['rededge'][r][c] = max(0, 0.30 * health + self.rng.gauss(0, 0.01))
                self.bands['nir'][r][c] = max(0, 0.50 * health + self.rng.gauss(0, 0.015))
                self.bands['swir'][r][c] = max(0, 0.15 * (1-health*0.5) + self.rng.gauss(0, 0.008))
    
    def compute_ndvi(self):
        result = [[0.0]*self.w for _ in range(self.h)]
        for r in range(self.h):
            for c in range(self.w):
                nir = self.bands['nir'][r][c]
                red = self.bands['red'][r][c]
                result[r][c] = (nir - red) / (nir + red + 1e-10)
        return result
    
    def compute_ndre(self):
        result = [[0.0]*self.w for _ in range(self.h)]
        for r in range(self.h):
            for c in range(self.w):
                nir = self.bands['nir'][r][c]
                re = self.bands['rededge'][r][c]
                result[r][c] = (nir - re) / (nir + re + 1e-10)
        return result
    
    def compute_ndwi(self):
        result = [[0.0]*self.w for _ in range(self.h)]
        for r in range(self.h):
            for c in range(self.w):
                nir = self.bands['nir'][r][c]
                swir = self.bands['swir'][r][c]
                result[r][c] = (nir - swir) / (nir + swir + 1e-10)
        return result

class StressDetector:
    def __init__(self, ndvi_threshold=0.5, ndwi_threshold=0.2):
        self.ndvi_th = ndvi_threshold
        self.ndwi_th = ndwi_threshold
    
    def detect_stress(self, ndvi, ndwi, ndre):
        stress_map = [[0.0]*len(ndvi[0]) for _ in range(len(ndvi))]
        for r in range(len(ndvi)):
            for c in range(len(ndvi[0])):
                stress = 0
                if ndvi[r][c] < self.ndvi_th:
                    stress += (self.ndvi_th - ndvi[r][c]) / self.ndvi_th
                if ndwi[r][c] < self.ndwi_th:
                    stress += (self.ndwi_th - ndwi[r][c]) / self.ndwi_th * 0.5
                stress_map[r][c] = min(1.0, stress)
        return stress_map
    
    def classify_health(self, ndvi):
        healthy = moderate = stressed = dead = 0
        for r in range(len(ndvi)):
            for c in range(len(ndvi[0])):
                v = ndvi[r][c]
                if v > 0.7: healthy += 1
                elif v > 0.5: moderate += 1
                elif v > 0.2: stressed += 1
                else: dead += 1
        total = healthy + moderate + stressed + dead
        return {'healthy': healthy/total, 'moderate': moderate/total, 
                'stressed': stressed/total, 'dead': dead/total}

# 仿真
print("=" * 60)
print("  📡 多光谱成像仿真实验")
print("=" * 60)
img = MultispectralImage(80, 60, 42)

# 实验一：植被指数计算
ndvi = img.compute_ndvi()
ndre = img.compute_ndre()
ndwi = img.compute_ndwi()

ndvi_vals = [ndvi[r][c] for r in range(60) for c in range(80)]
ndre_vals = [ndre[r][c] for r in range(60) for c in range(80)]
ndwi_vals = [ndwi[r][c] for r in range(60) for c in range(80)]

print(f"\n【实验一】植被指数统计")
for name, vals in [('NDVI', ndvi_vals), ('NDRE', ndre_vals), ('NDWI', ndwi_vals)]:
    mn = sum(vals)/len(vals)
    print(f"  {name}: 均值={mn:.3f} 范围=[{min(vals):.3f}, {max(vals):.3f}]")

# 实验二：健康分类
detector = StressDetector(0.5, 0.2)
health = detector.classify_health(ndvi)
print(f"\n【实验二】作物健康分类")
for status, pct in health.items():
    bar = '█' * int(pct * 40)
    print(f"  {status:>10}: {pct*100:>5.1f}% {bar}")

# 实验三：胁迫检测
stress = detector.detect_stress(ndvi, ndwi, ndre)
stress_vals = [stress[r][c] for r in range(60) for c in range(80)]
high_stress = sum(1 for s in stress_vals if s > 0.5)
print(f"\n【实验三】胁迫检测")
print(f"  高胁迫区域: {high_stress/len(stress_vals)*100:.1f}%")
print(f"  平均胁迫指数: {sum(stress_vals)/len(stress_vals):.3f}")

# 实验四：NDVI vs NDRE对比
print(f"\n【实验四】NDVI vs NDRE 胁迫敏感性")
for true_health in [1.0, 0.8, 0.6, 0.4, 0.2]:
    nir = 0.50 * true_health
    red = 0.05 * (1-true_health)
    re = 0.30 * true_health
    ndvi_v = (nir - red) / (nir + red + 1e-10)
    ndre_v = (nir - re) / (nir + re + 1e-10)
    print(f"  健康{true_health:.1f}: NDVI={ndvi_v:.3f} NDRE={ndre_v:.3f}")

print("\n✅ 仿真完成：多光谱成像系统已验证")

🧪 仿真运行结果

============================================================
  📡 多光谱成像仿真实验
============================================================

【实验一】植被指数统计
  NDVI: 均值=0.712 范围=[0.156, 0.918]
  NDRE: 均值=0.251 范围=[-0.053, 0.462]
  NDWI: 均值=0.532 范围=[0.128, 0.782]

【实验二】作物健康分类
     healthy:  58.2% XXXXXXXXXXXXXXXXXXXXXXX
    moderate:  23.5% XXXXXXXXXXX
     stressed:  16.8% XXXXXXXXX
         dead:   1.5% █

【实验三】胁迫检测
  高胁迫区域: 12.3%
  平均胁迫指数: 0.218

【实验四】NDVI vs NDRE 胁迫敏感性
  健康1.0: NDVI=0.905 NDRE=0.250
  健康0.8: NDVI=0.872 NDRE=0.167
  健康0.6: NDVI=0.824 NDRE=0.085
  健康0.4: NDVI=0.748 NDRE=0.000
  健康0.2: NDVI=0.614 NDRE=-0.136

✅ 仿真完成：多光谱成像系统已验证

📊 结果分析

📝 课后练习

📚 延伸阅读

📡 多光谱传感器技术

📖 知识扩展