不同降维策略下的高光谱影像多特征分类

由于物体表面的空间分布通常是富有规律且局部连续的,在高光谱影像分类中应充分利用其光谱和空间信息。本文在对高光谱影像立方体进行降维处理的基础上,提出了一种联合空域和谱域信息的高光谱影像高效分类方法。首先,分别选用主成分分析(Principal Component Analysis,PCA)和正交投影波段选择(Orthogonal Projection Band Selection,OPBS)两种方法对原始高光谱数据进行预处理,获取降维后的影像数据。然后在其基础上提取扩展形态学特征(Extended Morphology Profiles,EMP)和地物表面纹理特征,组成联合光谱和纹理、形状结构特征。最后,采用支持向量机(Support Vector Machine,SVM)分类器对联合特征进行分类。针对不同真实高光谱数据集的实验结果表明,本文提出的方法运算效率高且具有令人满意的分类性能。

Multi-feature Classification of Hyperspectral Images with Different Dimensionality Reduction Strategies

The spectral and spatial information should be fully utilized in the classification of hyperspectral images since the spatial distribution of the surface of the object is usually regular and locally continuous.In this paper,we proposed an efficient classification method for hyperspectral imagery based on the combination of spatial and spectral domain information.Firstly,Principal Component Analysis(PCA)and Orthogonal Projection Band Selection(OPBS)were used to pre-process the original hyperspectral data to obtain the reduced dimensional image data.On its basis,Extended Morphology Profiles(EMP)and surface texture features are extracted to form joint spectrum,texture,and shape structure features.Finally,the joint features are classified using the Support Vector Machine(SVM)classifier.Experimental results for different real hyperspectral datasets show that the proposed method has high computational efficiency and satisfactory classification performance.