论文部分内容阅读
自然界中不同种物质拥有不同的偏振特性,这些特征信号能用于检测不同的目标地物。为了探索偏振光谱技术用于精确识别作物和杂草的可行性,此研究利用配置偏振片的成像光谱仪FISS-P在室内采集玉米与5种杂草的偏振光谱影像。通过比较和分析0°、60°、120°和无偏4种状态下玉米与各种杂草的光谱响应规律、光谱特征和决策识别模型精度,结果显示4种偏振状态下玉米和杂草的光谱变化趋势较一致,无偏状态下玉米和杂草的光谱强度最大;不同偏振状态下玉米和杂草的敏感波段既存在共性又表现出一定的差异性;4种偏振状态下玉米杂草识别模型的总体精度和Kappa系数均达到90%以上,其中,0°偏振状态下玉米和杂草识别模型的整体精度最高,接近100%。综上,偏振光谱能够在叶片尺度较好地识别玉米和杂草,这为田间尺度进一步应用提供了扎实的数据积累。
Different substances in nature have different polarization characteristics, these characteristic signals can be used to detect different targets. In order to explore the feasibility of polarization spectroscopy for the accurate identification of crops and weeds, the polarization spectra of corn and five weeds were collected indoors using FISS-P, an imaging spectrometer equipped with a polarizer. By comparing and analyzing the spectral response laws, spectral characteristics and decision-making accuracy of maize and various weeds under 0 °, 60 °, 120 ° and unbiased conditions, the results show that the maize and weed The spectra of maize and weeds had the highest spectral intensity under unbiased condition. The sensitive bands of maize and weeds under different polarization states both showed commonness and showed some differences. The identification of maize weeds in four states of polarization The overall accuracy and the Kappa coefficient of the model are all over 90%. Among them, the overall accuracy of the corn and weed recognition model under 0 ° polarization state is the highest, which is close to 100%. In summary, the polarized spectrum can better identify corn and weeds at the leaf scale, which provides a solid data accumulation for further application in the field scale.