基于高光谱影像的SG滤波算法的研究

Research on SG Filtering Algorithm Based on Hyperspectral Image

高光谱遥感是当前遥感技术发展前沿,已经被成功应用于农业、水利、交通等许多领域.高光谱遥感影像中不仅存在空间域噪声而且存在光谱域噪声,传统的图像滤波只能对图像空间域噪声进行处理,而不能去除光谱域噪声.为了改变这种状况,本文提出了一种Savitzky-Golay(SG)滤波改进算法,诊断光谱域噪声.本文主要以航空高光谱遥感数据为研究对象,基于最小二乘的SG滤波求取反射率光谱的二阶导数,然后对影像进行噪声去除,该方法在保证有效去除光谱域噪声的同时,保留高光谱图像的大部分的细微特征.与其他不同的光谱域噪声滤波方法进行对比,实验证明本文的滤波方法是一种较为有效的手段.

Hyperspectral remote sensing is the forefront of the development of remote sensing technology,which has been successfully applied in agriculture,water conservancy,transportation and other fields. In hyperspectral remote sensing image,there is spectral noise as well as spatial noise. The traditional image filtering algorithm can deal with spatial noise,but can＇t remove spectral noise. In order to change this situation,an improved Savitzky-Golay（ SG）filtering algorithm is proposed to denoise it. In this paper,the aerial hyperspectral remote sensing data is taken as the research object, and the least square SG filter is used to obtain the second derivative of the reflectance spectrum,then noise removal of images is done. This method can effectively eliminate noise in spectral and retain most of the subtle features of hyperspectral images at the same time. Compared with other noise filtering methods,the experiment shows that the method is effective.