高光谱图像预处理方法研究及进展

The Research and Progress of Hyperspectral Image Preprocessing Method

高光谱图像预处理方法在高光谱图像处理中具有重要意义,有效的预处理方法可以尽可能地减少甚至消除无关信息(如样品背景、电噪音和杂散光等)对高光谱图像的影响,为后续基于高光谱图像的数据分析提供更为可靠的数据来源.基于高光谱图像谱图合一的数据结构特点,该文综述了直方图均衡化,中值滤波,边缘检测等图像预处理方法以及平滑,导数,标准归一化,多元散射校正等光谱预处理方法,并给出了这些方法的应用实例.该文还详细介绍了傅立叶变换和小波变换这两种基于数据压缩和信息提取的光谱预处理方法.这些新方法的研究为后续的数据分析奠定了良好的基础.

Hyperspectral image preprocessing method is significant in hyperspectral image pro‐cessing .Effective preprocessing method can reduce or even eliminate the bad effect caused by irrele‐vant information (such as background ,electrical noise and stray light etc .) ,and also provide more re‐liable data source for the subsequence analysis of hyperspectral image .Based on the special data struc‐ture characteristics integrating with spectrum and hyperspectral image ,the article summarizes the histogram equalization ,median filtering ,edge detection and the other image preprocessing methods commonly used ,such as smooth ,derivative ,normalization of standards ,the multiple scattering cor‐rection of spectral preprocessing methods ,meanwhile the article gives some examples of application of these methods .The article also introduces in detail the Fourier transform and wavelet transform based on the data compression and information extraction .The research of new preprocessing methods will lay the foundation for the subsequent data analysis of hyperspectral image .