可见光——近红外光谱矿物信息提取预处理技术

地物光谱特征研究是高光谱图像处理的基础,受观测环境的影响,野外测量的光谱数据中包含有噪声,不利于光谱矿物信息提取工作的进行.本文从光谱矿物信息提取的角度出发,探讨出一套光谱数据预处理流程.比较了小波降噪和Savitzky－Golay降噪的优缺点,实验表明,Savitzky－Golay方法更适合光谱降噪处理.在实际应用中,发现当前常用包络线算法存在一些局限性,在此基础上提出了一种改进的包络线算法.

Pretreatment Technology of Visible near Infrared Spectral Minerals Information Extraction

Research on spectral characteristics was the basis of hyperspectral image processing, due to the influence of the environment, spectral data contained the noisy, it was not conducive to the work of spectral minerals information extraction. From the viewpoint of spectrum minerals information extraction, this paper discussed a set of spectral data pretreatment process. Compared the advantages and disadvantages of the wavelet and Savitzky -- Golay noise reduction, and experiments showed that the Savitzky - Golay method was more suitable for spectrum noise reduction processing. In practice, found the current commonly envelope algorithm had some limitations, on this basis, gave out an improved envelope algorithm.