针对自适应波段选择法(adaptive band selection,ABS)对高光谱图像降维后得到的最优波段子集用于地物目标分类处理时,分类精度不理想的问题,提出一种K-means聚类与ABS结合的高光谱图像降维方法。算法采用K-means聚类算法对所有波段进行...针对自适应波段选择法(adaptive band selection,ABS)对高光谱图像降维后得到的最优波段子集用于地物目标分类处理时,分类精度不理想的问题,提出一种K-means聚类与ABS结合的高光谱图像降维方法。算法采用K-means聚类算法对所有波段进行聚类,聚类中分别采用相关系数和欧氏距离2种相似性度量,选取各聚类中ABS指数最大的波段,作为最优波段子集。通过实验,将所提方法与ABS进行分类精度比较。实验结果表明,所提方法在分类精度上优于ABS法,以相关系数作为相似性度量的K-means聚类与ABS结合的降维方法分类效果更好。展开更多
高光谱遥感数据能够提供比多光谱遥感数据更为丰富的光谱信息,从而更精确地刻画地物的光谱特征。在水体遥感原理基础上,采用自适应波段选择(adaptive band selection,ABS)方法对HJ-1A卫星高光谱数据的波段相关性和信息量进行分析,结合B...高光谱遥感数据能够提供比多光谱遥感数据更为丰富的光谱信息,从而更精确地刻画地物的光谱特征。在水体遥感原理基础上,采用自适应波段选择(adaptive band selection,ABS)方法对HJ-1A卫星高光谱数据的波段相关性和信息量进行分析,结合BP神经网络技术确定最优波段组合并构建盐湖矿物离子含量的反演模型,对柴达木盆地西台吉乃尔湖的K+,Mg2+,Na+,Cl-和SO2-4离子含量进行定量反演,获得盐湖矿物离子含量的空间分布情况。研究结果表明,BP神经网络反演模型的盐湖矿物离子含量反演精度在85%以上,反演得到的矿物离子含量的分布情况与实地调查结果基本一致。因此,利用高光谱数据和BP神经网络可以对盐湖矿物资源进行大范围动态监测,为盐湖资源的合理开发和高效利用提供科学依据。展开更多
高光谱数据波段多、波段之间相关性强,导致信息冗余严重,增加了数据处理的工作量,有效准确地在众多波段中选择具有代表性的波段尤为重要。首先用Wilks'Lambda(WL),随机森林(random forest,RF)与自适应波段选择(adaptive band select...高光谱数据波段多、波段之间相关性强,导致信息冗余严重,增加了数据处理的工作量,有效准确地在众多波段中选择具有代表性的波段尤为重要。首先用Wilks'Lambda(WL),随机森林(random forest,RF)与自适应波段选择(adaptive band selection,ABS)这3种方法对高光谱数据进行降维处理。然后提出了基于曲线误差指数的评价方法,用此指数的趋势来确定每种降维方法所要选择的合适波段数量,同时用指数的大小评价不同降维方法的优劣,并用分类方法对评价结果加以验证。结果显示:Wilks'Lambda最终选择的波段数为10个,α6-α平稳值(选择6个波段时的曲线误差值与曲线误差平稳值之间的差值)为0.05;随机森林最终选择的波段数为13个,α6-α平稳值为0.06;自适应波段选择方法最终选择的波段数为20个,α6-α平稳为0.14。Wilks'Lambda的总体分类精度为80.56%,Kappa系数为0.77;随机森林的总体分类精度为79.11%,Kappa系数为0.76;自适应波段选择方法的总体分类精度为49.94%,Kappa系数为0.41。得出以下结论:(1)基于曲线误差指数的方法得出Wilks'Lambda有最小的α6-α平稳值,是最佳的波段降维方法 ;分类结果显示:Wilks'Lambda有最大的总体分类精度与Kappa系数,是最佳的波段降维方法。(2)基于曲线误差指数的评价方法与基于分类结果的误差一致,说明此方法具有可行性。展开更多
Wavelet de-noising has been well known as an important method of signal de-noising. Recently,most of the research efforts about wavelet de-noising focus on how to select the threshold,where Donoho method is applied wi...Wavelet de-noising has been well known as an important method of signal de-noising. Recently,most of the research efforts about wavelet de-noising focus on how to select the threshold,where Donoho method is applied widely. Compared with traditional 2-band wavelet,3-band wavelet has advantages in many aspects. According to this theory,an adaptive signal de-noising method in 3-band wavelet domain based on nonparametric adaptive estimation is proposed. The experimental results show that in 3-band wavelet domain,the proposed method represents better characteristics than Donoho method in protecting detail and improving the signal-to-noise ratio of reconstruction signal.展开更多
Acoustic echo cancellation based on sub-band filters has the characteristics of rapid con- vergence and small computational complexity. This letter analyses two different sub-band filters design methods which used in ...Acoustic echo cancellation based on sub-band filters has the characteristics of rapid con- vergence and small computational complexity. This letter analyses two different sub-band filters design methods which used in acoustic echo cancellation fields and compares them with each other. Fur- thermore, the sub-band filter construction have been optimized, which lead to the improvement of the computational efficiency. At the same time, this letter combines ear auditory feature with acoustic echo cancellation, thus improves the original algorithms by importing a new objective function creatively. At the last part, a simulation environment has been designed and a computer simulation has been carried out. The final results indicate that this method can meet the requirements of actual projects, and some improvements are demonstrated on performance and calculation quantity compared to original algo- rithms.展开更多
文摘针对自适应波段选择法(adaptive band selection,ABS)对高光谱图像降维后得到的最优波段子集用于地物目标分类处理时,分类精度不理想的问题,提出一种K-means聚类与ABS结合的高光谱图像降维方法。算法采用K-means聚类算法对所有波段进行聚类,聚类中分别采用相关系数和欧氏距离2种相似性度量,选取各聚类中ABS指数最大的波段,作为最优波段子集。通过实验,将所提方法与ABS进行分类精度比较。实验结果表明,所提方法在分类精度上优于ABS法,以相关系数作为相似性度量的K-means聚类与ABS结合的降维方法分类效果更好。
文摘压缩是高光谱遥感(hyperspectral remote sensing)图像的一个重要研究领域.文中充分考虑了高光谱遥感图像的谱间相关性较强而空间相关性相对较弱的特点,采用了自适应波段选择降维方法与基于神经网络的矢量量化方法相结合的方法对高光谱遥感图像进行压缩.首先采用自适应波段选择(Adaptive band selection)的谱间压缩方法,通过自适应地选择信息量大并且与其他波段相关性小的波段来降低高光谱数据量.然后对降维后图像在空间进行小波变换并进行矢量量化,最后对量化后数据进行自适应算术编码.实验结果表明,谱间压缩能够保留信息丰富的波段,同时计算复杂度大大降低;基于神经网络的SOFM算法及其改进算法取得较好的空间压缩效果,实现了对高光谱遥感图像的有效压缩.
文摘高光谱遥感数据能够提供比多光谱遥感数据更为丰富的光谱信息,从而更精确地刻画地物的光谱特征。在水体遥感原理基础上,采用自适应波段选择(adaptive band selection,ABS)方法对HJ-1A卫星高光谱数据的波段相关性和信息量进行分析,结合BP神经网络技术确定最优波段组合并构建盐湖矿物离子含量的反演模型,对柴达木盆地西台吉乃尔湖的K+,Mg2+,Na+,Cl-和SO2-4离子含量进行定量反演,获得盐湖矿物离子含量的空间分布情况。研究结果表明,BP神经网络反演模型的盐湖矿物离子含量反演精度在85%以上,反演得到的矿物离子含量的分布情况与实地调查结果基本一致。因此,利用高光谱数据和BP神经网络可以对盐湖矿物资源进行大范围动态监测,为盐湖资源的合理开发和高效利用提供科学依据。
文摘高光谱数据波段多、波段之间相关性强,导致信息冗余严重,增加了数据处理的工作量,有效准确地在众多波段中选择具有代表性的波段尤为重要。首先用Wilks'Lambda(WL),随机森林(random forest,RF)与自适应波段选择(adaptive band selection,ABS)这3种方法对高光谱数据进行降维处理。然后提出了基于曲线误差指数的评价方法,用此指数的趋势来确定每种降维方法所要选择的合适波段数量,同时用指数的大小评价不同降维方法的优劣,并用分类方法对评价结果加以验证。结果显示:Wilks'Lambda最终选择的波段数为10个,α6-α平稳值(选择6个波段时的曲线误差值与曲线误差平稳值之间的差值)为0.05;随机森林最终选择的波段数为13个,α6-α平稳值为0.06;自适应波段选择方法最终选择的波段数为20个,α6-α平稳为0.14。Wilks'Lambda的总体分类精度为80.56%,Kappa系数为0.77;随机森林的总体分类精度为79.11%,Kappa系数为0.76;自适应波段选择方法的总体分类精度为49.94%,Kappa系数为0.41。得出以下结论:(1)基于曲线误差指数的方法得出Wilks'Lambda有最小的α6-α平稳值,是最佳的波段降维方法 ;分类结果显示:Wilks'Lambda有最大的总体分类精度与Kappa系数,是最佳的波段降维方法。(2)基于曲线误差指数的评价方法与基于分类结果的误差一致,说明此方法具有可行性。
基金the Scientific Research Foundation for the Returned Overseas Chinese Scholars of the Ministry of Education (No.2004.176.4)the Natural Science of Foundation Shandong Province (No.Z2004G01).
文摘Wavelet de-noising has been well known as an important method of signal de-noising. Recently,most of the research efforts about wavelet de-noising focus on how to select the threshold,where Donoho method is applied widely. Compared with traditional 2-band wavelet,3-band wavelet has advantages in many aspects. According to this theory,an adaptive signal de-noising method in 3-band wavelet domain based on nonparametric adaptive estimation is proposed. The experimental results show that in 3-band wavelet domain,the proposed method represents better characteristics than Donoho method in protecting detail and improving the signal-to-noise ratio of reconstruction signal.
文摘Acoustic echo cancellation based on sub-band filters has the characteristics of rapid con- vergence and small computational complexity. This letter analyses two different sub-band filters design methods which used in acoustic echo cancellation fields and compares them with each other. Fur- thermore, the sub-band filter construction have been optimized, which lead to the improvement of the computational efficiency. At the same time, this letter combines ear auditory feature with acoustic echo cancellation, thus improves the original algorithms by importing a new objective function creatively. At the last part, a simulation environment has been designed and a computer simulation has been carried out. The final results indicate that this method can meet the requirements of actual projects, and some improvements are demonstrated on performance and calculation quantity compared to original algo- rithms.