【摘要】 端元提取是高光谱图像混合像元分解的关键问题。针对正交子空间投影方法进行端元提取需要端元先验知识的问题,提出一种基于光谱最小信息熵的正交子空间投影高光谱图像端元提取方法。以光谱最小信息熵判定最优端元子集,同时将正交投影散度作为不同地物光谱向量之间相似性的测度指标用于判别端元。利用模拟数据和真实数据对算法进行验证,结果表明:算法不需先验知识,能够自动进行端元提取,且精度较高。更多还原

【Abstract】 Endmember extraction is the key problem of hyperspectral images mixed pixel decomposition. To address Orthogonal Subspace Projection method for endmember extraction needing to solve the problem of endmember prior knowledge, a hyperspectral image endmember extraction method of Orthogonal Subspace Projection based on spectral minimum shannon entropy was proposed. It determined the optimal endmembers subset using the criteria of spectral minimum shannon entropy,and the Orthogonal Projection Divergence as a measure index of similarity between different spectral vector for judging the endmember. The algorithm was verified by the simulated hyperspectral images and real hyperspectral images. Experimental results show that the proposed algorithm can extract endmember automatically without prior knowledge and achieve relatively high extraction accuracy.更多还原