在纹理丰富的高光谱图像中获得精确的噪声估计,是噪声估计任务中的难点。本文基于高光谱图像的空间规律性和光谱相关性,提出一种基于超像素分割的光谱去相关法。同质区域划分是许多噪声估计方法的关键步骤,精确的同质区域划分能有效提...在纹理丰富的高光谱图像中获得精确的噪声估计,是噪声估计任务中的难点。本文基于高光谱图像的空间规律性和光谱相关性,提出一种基于超像素分割的光谱去相关法。同质区域划分是许多噪声估计方法的关键步骤,精确的同质区域划分能有效提高噪声估计精度。为此,将简单线性迭代聚类算法(Simple linear iterative clustering algorithm,SLIC)与光谱-空间相似性结合,划分高光谱图像为局部结构相似的图像块,以保持同质特征;为了提高光谱间的区分能力,将光谱信息散度和光谱角联合作为光谱距离;结合多元线性回归在同质区域内去除光谱相关性,在获得的残差图上估计噪声水平。对不同地物复杂程度的模拟图像,添加不同程度的噪声,通过与多种方法比较,验证了本文方法的有效性和稳定性。最后,本文方法成功应用于Urban数据的噪声水平估计,准确识别出受噪声严重污染的波段。展开更多
An efficient despeclding algorithm is proposed based on stationary wavelet transform （SWT） for synthetic aperture radar （SAR） images. The statistical model of wavelet coefficients is analyzed and its performance i...An efficient despeclding algorithm is proposed based on stationary wavelet transform （SWT） for synthetic aperture radar （SAR） images. The statistical model of wavelet coefficients is analyzed and its performance is modeled with a mixture density of two zero-mean Gaussian distributions. A fuzzy shrinkage factor is derived based on the minimum mean square error （MMSE） criteria with Bayesian estimation. In the case above, the ideas of region division and fuzzy shrinkage arc adopted according to the interscale dependencies among wavelet coefficients. The noise-free wavelet coefficients are estimated accurately. Experimental results show that the algorithm proposed is superior to the refined Lee filter, wavelet soft thresbolding shrinkage and SWT shrinkage algorithms in terms of smoothing effects and edges preservation.展开更多
文摘在纹理丰富的高光谱图像中获得精确的噪声估计,是噪声估计任务中的难点。本文基于高光谱图像的空间规律性和光谱相关性,提出一种基于超像素分割的光谱去相关法。同质区域划分是许多噪声估计方法的关键步骤,精确的同质区域划分能有效提高噪声估计精度。为此,将简单线性迭代聚类算法(Simple linear iterative clustering algorithm,SLIC)与光谱-空间相似性结合,划分高光谱图像为局部结构相似的图像块,以保持同质特征;为了提高光谱间的区分能力,将光谱信息散度和光谱角联合作为光谱距离;结合多元线性回归在同质区域内去除光谱相关性,在获得的残差图上估计噪声水平。对不同地物复杂程度的模拟图像,添加不同程度的噪声,通过与多种方法比较,验证了本文方法的有效性和稳定性。最后,本文方法成功应用于Urban数据的噪声水平估计,准确识别出受噪声严重污染的波段。
基金A Postdoctoral Science Foundation of China (J63104020156) National Defence Foundation of China
文摘An efficient despeclding algorithm is proposed based on stationary wavelet transform （SWT） for synthetic aperture radar （SAR） images. The statistical model of wavelet coefficients is analyzed and its performance is modeled with a mixture density of two zero-mean Gaussian distributions. A fuzzy shrinkage factor is derived based on the minimum mean square error （MMSE） criteria with Bayesian estimation. In the case above, the ideas of region division and fuzzy shrinkage arc adopted according to the interscale dependencies among wavelet coefficients. The noise-free wavelet coefficients are estimated accurately. Experimental results show that the algorithm proposed is superior to the refined Lee filter, wavelet soft thresbolding shrinkage and SWT shrinkage algorithms in terms of smoothing effects and edges preservation.
