为提高高光谱遥感图像的分类精度,通过局部保留判别式分析与深度卷积神经网络(DCNN)算法,提出了基于局部保留降维卷积神经网络的高光谱图像分类算法。首先,用局部保留判别式分析对高光谱数据降维,再用二维Gabor滤波器对降维后的高光谱...为提高高光谱遥感图像的分类精度,通过局部保留判别式分析与深度卷积神经网络(DCNN)算法,提出了基于局部保留降维卷积神经网络的高光谱图像分类算法。首先,用局部保留判别式分析对高光谱数据降维,再用二维Gabor滤波器对降维后的高光谱数据进行滤波,生成空间隧道信息;其次,用卷积神经网络对原始高光谱数据进行特征提取,生成光谱隧道信息;再次,融合空间隧道信息与光谱隧道信息,形成空间-光谱特征信息,并将其输入到深度卷积神经网络,提取更加有效的特征;最后,采用双重优选分类器对最终提取的特征进行分类。将本文方法与CNN、PCA-SVM、CD-CNN和CNN-PPF等算法在Indian Pines、University of Pavia高光谱遥感数据库上进行性能比较。在Indian Pines、University of Pavia数据库上,本文算法识别的整体精度比传统CNN方法的整体精度分别高3. 81个百分点与6. 62个百分点。实验结果表明,本文算法无论在分类精度还是Kappa系数都优于另外4种算法。展开更多
为了提高高光谱遥感图像的分类精度,通过结合像元邻域谱与概率协同表示方法,提出了一种基于空间信息与光谱信息的分类方法.首先采用插值方法生成像元的邻域谱,然后用概率协同表示方法将待测样本进行分类.用所提出的方法在AVIRIS Indian ...为了提高高光谱遥感图像的分类精度,通过结合像元邻域谱与概率协同表示方法,提出了一种基于空间信息与光谱信息的分类方法.首先采用插值方法生成像元的邻域谱,然后用概率协同表示方法将待测样本进行分类.用所提出的方法在AVIRIS Indian Pines和Salinas scene高光谱遥感数据库上进行分类实验,并和主成分分析、支持向量机、稀疏表示分类器和协同表示分类器方法进行了比较.结果表明,所提出的方法在AVIRIS Indian Pines数据库上识别精度比主成分分析法高约17%,其识别精度和kappa系数都优于另外4种方法.该方法是一种较好的高光谱遥感图像分类方法.展开更多
A turbulent microstructure experiment was undertaken at a low latitude of 10°N in the South China Sea in late August 2012. The characteristics of the eddy diffusivity above 650 m were analyzed, which is one order...A turbulent microstructure experiment was undertaken at a low latitude of 10°N in the South China Sea in late August 2012. The characteristics of the eddy diffusivity above 650 m were analyzed, which is one order of magnitude larger than that in the open ocean at that low latitude. Enhanced eddy diffusivities by strong shears and sharp changes in topography were observed. The strongest eddy diffusivity occurred in the mixed layer, and it reached O(10^-2 m^2/s). Strong stratification in the thermocline inhibited the penetration of surface eddy diffusivities through the thermocline, where the mixing was weakest. Below the thermocline, where the background eddy diffusivity was approximately O(10^-6 m^2/s), the eddy diffusivity increased with depth, and its largest value was O(10^-3 m^2/s).展开更多
文摘为提高高光谱遥感图像的分类精度,通过局部保留判别式分析与深度卷积神经网络(DCNN)算法,提出了基于局部保留降维卷积神经网络的高光谱图像分类算法。首先,用局部保留判别式分析对高光谱数据降维,再用二维Gabor滤波器对降维后的高光谱数据进行滤波,生成空间隧道信息;其次,用卷积神经网络对原始高光谱数据进行特征提取,生成光谱隧道信息;再次,融合空间隧道信息与光谱隧道信息,形成空间-光谱特征信息,并将其输入到深度卷积神经网络,提取更加有效的特征;最后,采用双重优选分类器对最终提取的特征进行分类。将本文方法与CNN、PCA-SVM、CD-CNN和CNN-PPF等算法在Indian Pines、University of Pavia高光谱遥感数据库上进行性能比较。在Indian Pines、University of Pavia数据库上,本文算法识别的整体精度比传统CNN方法的整体精度分别高3. 81个百分点与6. 62个百分点。实验结果表明,本文算法无论在分类精度还是Kappa系数都优于另外4种算法。
文摘为了提高高光谱遥感图像的分类精度,通过结合像元邻域谱与概率协同表示方法,提出了一种基于空间信息与光谱信息的分类方法.首先采用插值方法生成像元的邻域谱,然后用概率协同表示方法将待测样本进行分类.用所提出的方法在AVIRIS Indian Pines和Salinas scene高光谱遥感数据库上进行分类实验,并和主成分分析、支持向量机、稀疏表示分类器和协同表示分类器方法进行了比较.结果表明,所提出的方法在AVIRIS Indian Pines数据库上识别精度比主成分分析法高约17%,其识别精度和kappa系数都优于另外4种方法.该方法是一种较好的高光谱遥感图像分类方法.
基金The "CAS/SAFEA International Partnership Program for Creative Research Teams" of Chinese Academy of Seiences under contract Nos XDA11010202,2013CB430303 and 41376022,41276021 and 41276023
文摘A turbulent microstructure experiment was undertaken at a low latitude of 10°N in the South China Sea in late August 2012. The characteristics of the eddy diffusivity above 650 m were analyzed, which is one order of magnitude larger than that in the open ocean at that low latitude. Enhanced eddy diffusivities by strong shears and sharp changes in topography were observed. The strongest eddy diffusivity occurred in the mixed layer, and it reached O(10^-2 m^2/s). Strong stratification in the thermocline inhibited the penetration of surface eddy diffusivities through the thermocline, where the mixing was weakest. Below the thermocline, where the background eddy diffusivity was approximately O(10^-6 m^2/s), the eddy diffusivity increased with depth, and its largest value was O(10^-3 m^2/s).