一种基于非线性增益小波滤波的高光谱影像去噪技术研究  

A New Hyperspectral Images Filtering Algorithm Based on Non-linear Wavelet Analyze

作　　者：苏俊英[1] 舒宁[1] 

机构地区：[1]武汉大学遥感信息工程学院,湖北武汉430079

出　　处：《遥感技术与应用》2008年第4期434-439,共6页Remote Sensing Technology and Application

基　　金：国家973计划资助项目(2006CB701303);国家自然科学基金资助项目(40371079)

摘　　要：提出了一种光谱域非线性增益的高光谱影像小波滤波去噪方法。首先对高光谱曲线进行小波变换,计算低频部分的方差并以此设定噪声和特征之间阈值,对属于噪声的高频系数置零,对属于细节特征的高频系数非线性增益,实现对高光谱曲线的滤波去噪。通过与高光谱影像均值滤波平滑法、最小二乘平滑法对比分析结果表明,该方法在噪声抑制和细节保持方面取得了较好的效果。A non-linear strength wavelet filtering algorithm is proposed to remove noise from the hyper spectral images. First, the spectral curve is decomposed into high frequency coefficient and low frequency coefficient with wavelet transformation. Deviation is calculated with low frequency parts to determine the threshold of noise and detail feature. A non-linear filtering algorithm is proposed to deal with the noise and detail feature of high frequency coefficient with the deviation threshold of low frequency coefficient. Comparing with the result of mean filter smoothing and least squares smoothing, the non-linear strength wavelet filtering algorithm （nLWF） can obtain better effect in noise remove and detail feature conversation.

关 键 词：高光谱影像 小波滤波 光谱曲线 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]