基于PCA和SVM的高光谱遥感图像分类研究

Study on the hyperspectral remote sensed image classify based on PCA and SVM

支持向量机(SVM)是根据统计学习理论提出的新的研究方法,它在解决小样本、非线性及高维模式识别问题中表现出了许多特有的优势,在模式识别、函数逼近和概率密度估计等方面取得了良好的效果。由于高光谱图像波段数目多,各波段间具有较强的相关性,因此通过主成分分析(PCA)方法对高光谱数据进行预处理,达到了降维的目的,同时也去除了噪声波段。用支持向量机方法对高光谱遥感图像进行分类,可实现图像的分类识别。

SVM(Support Vector Machine) is a new researching method based on statistics theory.It has many advantages when solving small sample,nonlinear and high dimension model identification problems,so it acquires good effect in aspect of model identification,function approximation and probability density estimation,etc.Because hyperspectral image has lots of bands which have strong relevance,the hyperspectral data are pretreated by PCA(Principal Components Analysis) to decrease the dimension,simultaneously remove noise bands.Using SVM realizes classification and identification of hyperspectral remote sensed image.