高光谱图像混合噪声去除是遥感领域的一个基本问题,也是一个重要的预处理步骤。本研究针对高光谱图像去噪问题,为有效地对高光谱图像进行恢复,提出了一种基于重叠组稀疏性超拉普拉斯正则化(OGS-HL)的新型去噪方法。该方法可以有效捕捉...高光谱图像混合噪声去除是遥感领域的一个基本问题,也是一个重要的预处理步骤。本研究针对高光谱图像去噪问题,为有效地对高光谱图像进行恢复,提出了一种基于重叠组稀疏性超拉普拉斯正则化(OGS-HL)的新型去噪方法。该方法可以有效捕捉图像的局部相关性和方向性结构,同时减少传统全变分正则化中的阶梯伪影。通过乘子交替方向法求解非凸优化问题,显著提高了去噪效率。在多个遥感图像数据集上的仿真实验表明,所提方法在峰值信噪比(PSNR)和结构相似度(SSIM)等评价指标上优于现有技术,展现了在复杂噪声环境下的优越去噪性能和广泛的应用潜力。The removal of mixed noise from hyperspectral images is a fundamental issue in the field of remote sensing and an important preprocessing step. This study focuses on the denoising problem of hyperspectral images. To effectively restore hyperspectral images, a new denoising method based on Overlap Group Sparse Hyper Laplacian Regularization (OGS-HL) is proposed. This method can effectively capture the local correlation and directional structure of images, while reducing the step artifacts in traditional total variation regularization. By using the alternating direction method of multipliers to solve non-convex optimization problems, the denoising efficiency has been significantly improved. Simulation experiments on multiple remote sensing image datasets have shown that the proposed method outperforms existing technologies in evaluation metrics such as peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), demonstrating superior denoising performance and broad application potential in complex noisy environments.展开更多
提出了一种基于核主成分分析(kernel principal component analysis,简称KPCA)和拉普拉斯正则化最小二乘(Laplacian regularized least squares,简称LapRLS)的合成孔径雷达(synthetic aperture radar,简称SAR)目标识别方法.KPCA特征提...提出了一种基于核主成分分析(kernel principal component analysis,简称KPCA)和拉普拉斯正则化最小二乘(Laplacian regularized least squares,简称LapRLS)的合成孔径雷达(synthetic aperture radar,简称SAR)目标识别方法.KPCA特征提取方法不仅能够提取目标主要特征,而且有效地降低了特征维数.Laplacian正则化最小二乘分类是一种半监督学习方法,将训练集样本作为有标识样本,测试集样本作为无标识样本,在学习过程中将测试集样本包含进来以获得更高的识别率.在MSTAR实测SAR地面目标数据上进行实验,结果表明,该方法具有较高的识别率,并对目标角度间隔具有鲁棒性.与模板匹配法、支撑矢量机以及正则化最小二乘监督学习方法相比,具有更高的SAR目标识别正确率.此外,还通过实验分析了不同情况下有标识样本数目对目标识别性能的影响.展开更多
Semi-supervised learning is an emerging computational paradigm for machine learning,that aims to make better use of large amounts of inexpensive unlabeled data to improve the learning performance.While various methods...Semi-supervised learning is an emerging computational paradigm for machine learning,that aims to make better use of large amounts of inexpensive unlabeled data to improve the learning performance.While various methods have been proposed based on different intuitions,the crucial issue of generalization performance is still poorly understood.In this paper,we investigate the convergence property of the Laplacian regularized least squares regression,a semi-supervised learning algorithm based on manifold regularization.Moreover,the improvement of error bounds in terms of the number of labeled and unlabeled data is presented for the first time as far as we know.The convergence rate depends on the approximation property and the capacity of the reproducing kernel Hilbert space measured by covering numbers.Some new techniques are exploited for the analysis since an extra regularizer is introduced.展开更多
文摘高光谱图像混合噪声去除是遥感领域的一个基本问题,也是一个重要的预处理步骤。本研究针对高光谱图像去噪问题,为有效地对高光谱图像进行恢复,提出了一种基于重叠组稀疏性超拉普拉斯正则化(OGS-HL)的新型去噪方法。该方法可以有效捕捉图像的局部相关性和方向性结构,同时减少传统全变分正则化中的阶梯伪影。通过乘子交替方向法求解非凸优化问题,显著提高了去噪效率。在多个遥感图像数据集上的仿真实验表明,所提方法在峰值信噪比(PSNR)和结构相似度(SSIM)等评价指标上优于现有技术,展现了在复杂噪声环境下的优越去噪性能和广泛的应用潜力。The removal of mixed noise from hyperspectral images is a fundamental issue in the field of remote sensing and an important preprocessing step. This study focuses on the denoising problem of hyperspectral images. To effectively restore hyperspectral images, a new denoising method based on Overlap Group Sparse Hyper Laplacian Regularization (OGS-HL) is proposed. This method can effectively capture the local correlation and directional structure of images, while reducing the step artifacts in traditional total variation regularization. By using the alternating direction method of multipliers to solve non-convex optimization problems, the denoising efficiency has been significantly improved. Simulation experiments on multiple remote sensing image datasets have shown that the proposed method outperforms existing technologies in evaluation metrics such as peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), demonstrating superior denoising performance and broad application potential in complex noisy environments.
文摘提出了一种基于核主成分分析(kernel principal component analysis,简称KPCA)和拉普拉斯正则化最小二乘(Laplacian regularized least squares,简称LapRLS)的合成孔径雷达(synthetic aperture radar,简称SAR)目标识别方法.KPCA特征提取方法不仅能够提取目标主要特征,而且有效地降低了特征维数.Laplacian正则化最小二乘分类是一种半监督学习方法,将训练集样本作为有标识样本,测试集样本作为无标识样本,在学习过程中将测试集样本包含进来以获得更高的识别率.在MSTAR实测SAR地面目标数据上进行实验,结果表明,该方法具有较高的识别率,并对目标角度间隔具有鲁棒性.与模板匹配法、支撑矢量机以及正则化最小二乘监督学习方法相比,具有更高的SAR目标识别正确率.此外,还通过实验分析了不同情况下有标识样本数目对目标识别性能的影响.
基金supported by National Natural Science Foundation of China (Grant Nos.11171014 and 11101024)National Basic Research Program of China (973 Project) (Grant No. 2010CB731900)
文摘Semi-supervised learning is an emerging computational paradigm for machine learning,that aims to make better use of large amounts of inexpensive unlabeled data to improve the learning performance.While various methods have been proposed based on different intuitions,the crucial issue of generalization performance is still poorly understood.In this paper,we investigate the convergence property of the Laplacian regularized least squares regression,a semi-supervised learning algorithm based on manifold regularization.Moreover,the improvement of error bounds in terms of the number of labeled and unlabeled data is presented for the first time as far as we know.The convergence rate depends on the approximation property and the capacity of the reproducing kernel Hilbert space measured by covering numbers.Some new techniques are exploited for the analysis since an extra regularizer is introduced.
