基于RICA的高光谱遥感影像地物识别研究

Research on land object recognition of hyperspectral remote sensing image based on RICA

高光谱影像具有光谱分辨率高、光谱覆盖范围更宽且波段多等优点,但其数据量巨大导致数据存在大量冗余、图像数据预处理困难,极大地限制了高光谱影像的应用。文章在ICA算法的基础上,以软重建惩罚替换ICA的正交性约束来克服ICA的缺点,实现基于RICA的高光谱降维算法。本研究对高分5号遥感影像进行PCA、ICA和RICA降维实验,应用支持向量机进行分类。实验结果表明,RICA算法在高光谱降维中是更加有效的。

Hyperspectral images have the advantages of high spectral resolution, wider spectral coverage and multiple bands. However, the huge amount of data brings about a lot of redundancy in the data, and the difficulty of image data preprocessing is also very obvious. This limits the application of hyperspectral imaging. On the basis of ICA algorithm,this paper replaces the orthogonality constraint of ICA with soft reconstruction penalty to overcome the shortcomings of ICA, and realizes the hyperspectral dimensionality reduction algorithm based on RICA. The PCA, ICA and RICA dimensionality reduction experiments were carried out on the remote sensing image of Gaofen-5, and the support vector machine was used for classification. The experimental results show that the RICA algorithm is more effective in hyperspectral dimensionality reduction.