多端元光谱解混模型的改进及对植被盖度的反演

摘要

以扎龙自然保护区为研究对象,运用分层的面向对象分类法与多端元光谱解混算法相结合反演该地区的植被覆盖度.结果表明:分层降低了场景复杂度,面向对象分类法与多端元光谱解混算法的结合,有效的减少了计算量和混合像元的端元变化;采用同期高分辨率的SPOT5多光谱遥感影像进行精度验证,与传统的多端元光谱解混模型的反演结果进行对比,相关系数从0.8643提高到0.9028,均方根误差从0.1712减少到0.0926.因此,分层面向对象多端元光谱解混模型适合对湿地植被覆盖度的反演.%In Zhalong Nature Reserve, the vegetation coverage of the area was retrieved by combining hierarchical object-oriented classification with Multiple Endmember Spectral Mixture Analysis ( MESMA) .Stratification reduced the complexity of the scene, the combination of object-oriented classification and MESMA effectively reduced computation and endmember varia-tion of mixed pixels.A high resolution SPOT5 multi spectral remote sensing image was used to verify the accuracy of the same time.Compared with the inversion results of the traditional MESMA, the correlation coefficient was increased from 0. 8643 to 0.9028, and the root mean square error was decreased from 0.1712 to 0.0926.Therefore, the hierarchical object-oriented Multiple Endmember Spectral Mixture Model is suitable for the inversion of the vegetation coverage of the wetland.