高光谱空谱一体化图像分类研究

Classification of hyperspectral remote sensing image based on spatial-spectral integration

高光谱图像分类是遥感图像处理技术中的一个热点,提高分类精度是目前一个重要研究方向。常规的高光谱图像分类技术主要关注于如何更好地利用光谱空间的分类信息,往往忽视图像空间域信息。本文提出了一种基于空谱一体化处理的高光谱图像分类方法,在利用数据进行自身光谱特征分类的同时采用区域生长法和二值形态学法相结合的空间域有效信息对光谱分类结果进行补充。实验证明本方法能提高高光谱图像分类精度。

Hyperspectral image classification is an important research area in remote sensing data processing, and ex-tensive research has been carried out to obtain higher classification accuracy. The traditional hyperspectral image clas-sification techniques usually concentrate on the information drawn from the spectral domain, while the information of spatial domain is ignored. In this paper, a hyperspectral classification method based on the combination of spectral and spatial information is proposed. Spatial domain methods, such as the region growing method and the binary mor- phology method,are applied to complement the classification result from the spectral domain information. Experimen-tal results based on a hyperspectral data set show that the proposed method has the capability to increase the classifi- cation accuracy.