基于噪声白化的高光谱数据子空间维数算法

发布时间：2018-06-15 16:45

本文选题：高光谱降维 + HySime　；参考：《国土资源遥感》2017年02期

【摘要】：高光谱影像数据的相邻波段间相关性较强,信号与噪声共存,根据最小二乘原理,使观测数据与噪声的投影误差之和最小化的HySime(hyperspectral signal identification by minimum error)算法,通过数据观测值减去噪声估计值后得到信号的估计值,进而可以计算信号相关矩阵的估计值。该算法在准确估计噪声的情况下是可行的,但实际上经光谱降维去相关后得到的各像元噪声估计值往往并不准确,因此,原始的HySime算法得到的结果可能并不理想。提出一种基于噪声白化的HySime改进算法,它不必进行逐像元的噪声去除,而是先对原始数据进行噪声白化处理,然后准确获取噪声的协方差矩阵估计值,再利用HySime算法进行信号相关矩阵计算,实现了提高算法精度的目的。通过模拟和实验数据的验证,改进的算法结果更准确稳定,与经典的NSP(noise subspace projection)算法在不同情况下所得结果有很好的一致性,通过引入噪声白化的过程,提高了算法对非白噪声的适应性。
[Abstract]:The correlation between adjacent bands of hyperspectral image data is strong, and the signal and noise coexist. According to the principle of least squares, the HySimeon hyperspectral signal identification by minimum error) algorithm, which minimizes the sum of projection errors between the observed data and the noise, is proposed. The estimated value of the signal can be obtained by subtracting the estimated value of the noise from the data observation value, and the estimation value of the signal correlation matrix can be calculated. This algorithm is feasible in the case of accurate noise estimation, but in fact, the pixel noise estimation obtained by spectral dimensionality de-correlation is often not accurate. Therefore, the original HySime algorithm may not be satisfactory. An improved HySime algorithm based on noise whitening is proposed. It does not have to remove the noise from pixel by pixel, but firstly whitens the original data, and then accurately obtains the estimated value of the covariance matrix of noise. Then HySime algorithm is used to calculate the signal correlation matrix and the precision of the algorithm is improved. The results of simulation and experimental data show that the improved algorithm is more accurate and stable, which is in good agreement with the results obtained by the classical NSPP noise subspace projection algorithm in different cases, and the process of noise whitening is introduced. The adaptability of the algorithm to non-white noise is improved.
【作者单位】：中国国土资源航空物探遥感中心;中国科学院遥感与数字地球研究所;
【基金】：中国地质调查局地质调查项目“天山-北山重要成矿区带遥感调查”(编号:121201003000150008);“高光谱地质调查技术方法研究”(编号:1212031513012)共同资助
【分类号】：TP751

【相似文献】