提出对多光谱数据进行对数变换来凸显类型特征,然后进行主成分变换并根据主成分贡献率确定EM算法分类所需主成分数,消除方差协方差矩阵的奇异性,同时削弱噪声;对数变换后的第一主成分直方图充分反映类型信息,由此确定的初始类别标签作为多个主成分EM分类算法所需初始值,避开随机选初值的敏感问题。实验证明,所提出的计算方案分类精度优于普通EM方法和传统的K-means方法。 更多还原
【Abstract】 It is proposed that the class differences are emphasized by firstly logarithmizing the original data,then apply the principal component transformation to the modified data.The number of principal components for the EM algorithm is determined according to the contribution rate(>85%).The proposed method not only removes the singularity but weakens the noise.The histogram of the first principal component sufficiently reflects the class differences,from which the initial label for multi-dimension... 更多还原