基于核方法的光谱角制图模型及其在高光谱图像分割中的应用

Kernel Method Based Spectral Angle Mapping Model and Its Applications to Hyperspectral Image Segmentation

在分析传统光谱角制图方法的几何属性和相似性度量特点的基础上,提出一种新的光谱相似性度量方法。该方法首先利用非线性变换将原始光谱向量变换到一个新的特征空间。其次,在特征空间中,利用协方差矩阵的非零特征向量构造了一组正交基,并将变换后的光谱向量投影到特征空间中的正交基上。高光谱图像分割实验结果表明该方法在光谱相似性度量上优于传统的光谱角制图方法。

Based on the analysis of geometric attributes and characteristic of spectral angle mapping, a novel similarity measure is presented in this paper. In this method, all original spectral vectors are, firstly, nonlinearly transformed into a feature space. Next, kernel PCA is used to construct a set of orthogonal coordinate base in feature space. All transformed spectral vectors are projected onto the orthogonal coordinate space. At last, spectral angle mapping method are employed to measure the similarity between two spectra with some added restrictions. Experimental results of Hyperspectral image segmentation show that our method is better than traditional spectral angle mapping method.