论文部分内容阅读
特征提取、度量与分析是高光谱遥感应用的基础.面向高光谱遥感数据的特点,将光谱特征划分为点尺度、面尺度和体尺度三个尺度的特征.基于特征属性与算法原理,构建了光谱曲线特征、光谱变换特征和光谱度量特征三个层次的高光谱遥感光谱特征体系,并对光谱特征提取与应用进行了深入探讨.光谱曲线特征包括直接光谱编码、光谱反射与吸收特征等,光谱变换特征包括植被指数、导数光谱等,光谱度量特征则包括光谱角、SID、相关系数和距离等.在分析特征算法原理的同时对其特点和应用进行了探讨.试验表明四值编码、光谱角和SID在应用中能够取得较好的效果.
Feature extraction, measurement and analysis are the foundation for the application of hyperspectral remote sensing.According to the characteristics of hyperspectral remote sensing data, the spectral features are divided into three scales: point scale, surface scale and body scale.According to the characteristics of attributes and algorithm principle, Spectral curve characteristics, spectral transformation characteristics and spectral measurement characteristics of the three levels of hyperspectral remote sensing spectral characteristics of the system, and spectral characteristics of the extraction and application of in-depth.Coupled spectral characteristics include direct spectral coding, spectral reflection and absorption characteristics, the spectrum The transformation features include vegetation index, derivative spectrum, etc. Spectral measurement features include spectral angle, SID, correlation coefficient and distance, etc. The characteristics and applications of the algorithm are discussed in detail.Experiments show that the four-value encoding, spectral angle And SID in the application can achieve better results.