摘要
混合光谱分解研究对提高遥感识别地物信息精度具有重要意义。实验室模拟混合光谱分解模型可以作为遥感影像混合像元分解的基础,为地物分类提供经验模型和理论依据。对土壤-植被混合模型进行实验室模拟,对原始光谱反射率进行一阶微分、对数、去连续统变换,运用特征波段法和相似系数法拟合回归方程,并对比模型精度。结果表明基于对数变换数据的特征波段法和一阶导数变换数据的相似系数法建模,反演组分含量的精度高。
Study of spectral unmixing plays an important role in increasing the accuracy of surface features information identifying using remote sensing,spectral unmixing model of laboratory simulation can be used as the basis of remote sensing image unmixing,providing classification models and theoretical basis for terrain classification.The mixing model with soil-plant leaves,conduct a first order derivative to the original reflectance is simulatee,taken logarithm for the original,and remove continuum,used characteristic bands and similarity coefficient to write regression equation,and compared the accuracy of each model.The results demonstrate that high accuracy on inversing components content can be acquired through two different methods.The first one is taking logarithm for the original reflectance dataset first,and then applying characteristic bands to write equation.The second method is based on conducting the first order derivative to the original reflectance with equation written by similarity coefficient.
出处
《科学技术与工程》
2011年第35期8785-8790,共6页
Science Technology and Engineering
基金
中国地质调查局(1212010761502)资助
