论文部分内容阅读
针对现有端元提取方法从数据光谱或空间或特征信息的单一方面出发进行混合像元分解、不同类型端元难以区分等不足,提出一种扩展形态学与正交子空间投影结合的端元自动提取方法.利用扩展膨胀和腐蚀操作,通过计算形态离心率指数进行高光谱数据的端元数据集计算;利用光谱角匹配方法提取不同类型的端元,通过向端元正交子空间投影消除已经提取端元的影响;并通过航空高光谱数据进行算法性能验证.实验结果表明:提出方法能够实现在无任何先验信息情况下图像端元的自动提取,并且能够有效地区分相似光谱端元.
Aiming at the problem that the existing endmember extraction method is based on the single aspect of spectrum or space or feature information of the data, it is difficult to distinguish the different types of endpoints from the mixed pixel decomposition. A new method is proposed to combine the extended morphology with the orthogonal subspace projection Element automatic extraction method.Using expansion expansion and erosion operation, the end-point data set of hyperspectral data is calculated by calculating the morphological eccentricity index.Different types of end elements are extracted by the spectral angle matching method, and by projecting to the end-element orthogonal subspace And the performance of the algorithm has been verified by using the hyperspectral data.The experimental results show that the proposed method can automatically extract the image endpoints without any prior information and can effectively distinguish the similar spectral end yuan.