Research On Dimensional Reduction Method Of Hyperspectral Remote Sensing Images

Hyperspectral remote sensing was founded on electrical-magnetic spectral theory, ground rule, electrical technology, computer science and spatial technology, it is developed rapidly as a new independent integrated technology. Because of the special high resolution of hyperspectral remote sensing, its latent usability has been paid attention on. Many methods have been researched aiming at multispectral image, which tend to perfect. But the large data and high dimensions of hyperspectral image disabled implementing methods of multispectral into hyperspectral directly, so it is important to explore methods of hyperspectral image. This paper put its emphasis on dimension reduction, the main research contents are as following:The characteristic of hyperspectral remote sensing image is researched. Prove hyperspectral remote sensing image relatively stronger spectrum dependence and relatively weaker space dependence.This paper researches lowering dimensionalities of hyper-spectral image based on Principle Component Analysis (PCA). With eigenvalues and eigenvectors of the covariance metrics of the original data, the contribution of principle component can be calculated, the sum of the contribution of all important principle components can reflect the information, which can be used to feature extraction, the results of the experiments indicate that the algorithm can get good results.This paper researches lowering dimensionalities of hyper-spectral image based o on a hybrid wavelet-PCA. PCA is sufficient for reducing data volume, however it fails to preserve the spectral and local characteristic of the original data. Wavelet decomposition which can preserve the spectral and local characteristic focuses on reducing each individual pixel in the spectral domain, this algorithm makes full use of the advantages of PCA and wavelet decomposition, which first performs an initial reduction using a wavelet decomposition, and then the feature extraction is applied. The results of the experiments indicate the algorithm has its superiority.This paper researches lowering dimensionalities of hyper-spectral image based on Kernel Principal Component Analysis. The principal component analysis only refers to second-order statistic information of image data and does not utilize the high-order ones. It ignores its nonlinear relativity among a large number of pixels. Research indicates that high-order statistic information sometimes contains image edge or nonlinear relations among multi-pixels. Aiming at this problem, this paper proposes kernel method and studies all its primary components. Then, we combine kernel method and principal component analysis to reduce the dimensionality of the hyperspectral image. The results of the experiments indicate the algorithm can preferably reserve information of hyperspectral image.