一种结合稀疏表示和纹理分块的遥感影像超分辨率方法

提出了一种基于稀疏表示和纹理分块的单幅遥感影像超分辨率方法,主要利用先验知识及影像自身的纹理信息重构遥感图像。首先,提取用于字典学习的图像块,从高、低分辨率遥感图像块中训练出冗余字典,采用正交匹配追踪方法更新字典,用迭代的方法直到算法收敛;然后,将训练的字典应用于遥感影像超分辨率重构。重构时将图像块分成平滑块和非平滑块两种类型,平滑块采用双三次卷积方法重构,非平滑块采用低分辨率遥感图像块的稀疏表示系数及高分辨率图像块冗余字典重构。实验结果表明,此方法重构速度较快,并在视觉及客观评价指标上有较好的超分辨率效果。     

Improved Image Pansharpening Technique using Nonsubsampled Contourlet Transform with Sparse Representation

Multispectral (MS) and panchromatic (PAN) images contains complementary information. High spatial and spectral resolution is a prerequisite for images to be useful, which can be achieved through image pansharpening. In this paper, we propose a new pansharpening technique which is a combination of nonsubsampled contourlet transform (NSCT) and sparse representation (SR), called NSCT–SR. NSCT is a shift-invariant version of the contourlet transform which combines nonsubsampled pyramid (NSP) and the directional filter banks. NSP splits input MS and PAN images into low-pass and high-pass sub-bands. Fusion of high-pass sub-bands is done using local energy information while low-pass sub-bands are fused using SR. Finally, fused low-pass and high-pass sub-bands are combined to obtain image with high spatial and high spectral resolution. We have quantitatively compared NSCT–SR with other multiresolution algorithms by calculating spatial and spectral quality parameters. It is observed that spatial quality is improved by 0.93 % (for seaside image) and 1.54 % (for urban image). While spectral quality is improved maximum up to 31.39 and 40.47 %, for respective images. NSCT–SR also compared with other state-of-art algorithms by calculating various performance parameters including quality with no reference. It is found that, overall; NSCT–SR performs better compared to algorithms considered in work.     

联合显著性和多层卷积神经网络的高分影像场景分类

高分辨率遥感影像中的场景信息,对于影像解译和现实世界的理解具有重要意义。传统的场景分类方法多利用中、低层人工特征,但是高分辨率遥感影像的信息丰富,场景构成复杂,需要高层次的特征来表达。本文提出了一种联合显著性和多层卷积神经网络的方法,首先利用显著性采样获取包含影像主要信息的有意义的块,将这些块作为样本集输入卷积神经网络中进行训练,获得不同层次的特征表达,最后联合多层特征利用支持向量机进行分类。两组高分影像场景数据UC Merced 21类和Wuhan 7类试验表明,显著性采样能够有效地获取主要目标,减弱其他无关目标的影响,降低数据冗余;卷积神经网络能够自动学习高层次的特征,相比已有方法,本文方法能够有效提高分类精度。     

利用在线字典学习实现图像超分辨率重建的算法

图像超分辨率重建是通过对单张或多张具有互补信息的低分辨率图像进行处理,重建一张高分辨率图像的技术。在单张图像的超分辨率重建中,基于稀疏表示的方法取得了很好的效果,得到了广泛的应用。一张图像中不同区域的图像块的内容一般会有显著变化。而基于稀疏表示的超分辨率重建算法多采用固定的字典,无法适应每一个图像块的重建需求。提出了一种结合外部数据和输入图像自身信息进行超分辨率重建的方法,通过搜索待处理图像块的非局部自相似块,结合在线字典学习方法对字典进行更新,从而保证更新后的字典能够匹配待处理的图像块。采用包括遥感图像在内的5张图像进行实验,并与4种经典的超分辨率重建算法进行比较,实验结果表明,此算法在主观评价和客观评价方面都有更好的表现。     

主题学习和稀疏表示的MODIS图像超分辨率重建

针对MODIS图像分辨率受传感器限制和噪声干扰,且分辨率局限在一定水平等问题,提出一种采用主题学习和稀疏表示的MODIS图像超分辨率重建方法,该方法通过双边滤波将MODIS图像的平滑及纹理部分分离,并将纹理部分看成是由若干"文档"组成的训练样本;运用概率潜在语义分析提取"文档"的潜在语义特征,从而确定"文档"所属的"主题"。在此基础上,针对每个主题所对应的图像块,采用改进的K-SVD方法训练若干适用于不同主题的高低分辨率字典对,从而可以运用这些字典对,通过稀疏编码实现测试图像相应主题块的超分辨率重建。实验结果表明,重建图像在视觉效果和PSNR等指标上均优于传统方法。     

利用多尺度特征与深度网络对遥感影像进行场景分类

针对因样本量小而导致的遥感图像场景分类精度不高的问题,结合非下采样Contourlet变换(NSCT)、深度卷积神经网络(DCNN)和多核支持向量机(MKSVM),提出了一种基于多尺度深度卷积神经网络(MS-DCNN)的遥感图像场景分类方法。首先利用非下采样Contourlet变换方法对遥感图像多尺度分解,然后对分解后的高频子带和低频子带分别用DCNN训练得到了不同尺度的图像特征,最后采用MKSVM综合多尺度特征并实现遥感图像场景分类。对标准遥感图像分类数据集的试验结果表明,本算法能够结合低频和高频子带对不同类别场景的识别优势,对遥感图像场景取得较好的分类结果。     

Improved IHS Algorithm for Fusing High Resolution Satellite Images of Urban Areas

The intensity-hue-saturation method is used frequently in image fusion due to its efficiency and high spatial quality. The main shortage is its spectral distortion stemmed from replacement of intensity band with higher resolution image. In this study, a new method is introduced to improve the spectral quality of the Intensity-Hue-Saturation (IHS) algorithm. The goal of this study is to produce the fused image that has a better spectral and spatial quality with respect to the original images in term of visual comparison and the classification result. In this regard, an improved statistical approach is developed to combine an intensity band from IHS algorithm and an input high resolution image such as SAR or Panchromatic image. Then the intensity image is replaced by the combined image band. Final fused images are attained using the inverse IHS algorithm. The proposed fusion algorithm is tested on two data sets of: a) panchromatic and multi spectral bands of IKONOS image with the same acquisition date, and b) multi spectral and HH bands of IKONOS and TerraSAR-X images respectively with different acquisition dates. Moreover, the obtained results are compared with other fusion methods like IHS, Gungor, Brovey and synthetic variable ratio. The results show less spectral discrepancy of the proposed method comparing to other methods. Finally, the outcome of proposed method is classified and classification overall accuracy is improved by 5.6 and 2 percentage for data set ‘a’ and ‘b’ respectively.     

基于稀疏表示和字典学习的全色与多光谱影像融合

Pansharpening方法通过融合多光谱影像的光谱信息和全色影像的空间细节信息来得到高分辨多光谱影像。然而传统的Pansharpening方法易导致产生光谱扭曲和空间信息丢失现象。受到影像稀疏表示超分重建理论启发,本文提出了一种新的基于稀疏表示和字典学习的Pansharpening方法。该方法以影像的高频特征作为训练样本,通过字典学习的方法来获取高低分辨率影像字典,使用正交匹配追踪算法求解出影像的稀疏表示系数,最终通过高分辨影像字典与稀疏系数相乘得到融合影像。实验结果表明:本文提出的方法能很好地保持遥感影像的光谱信息和空间细节信息。     

一种顾及区域特征差异的热红外与可见光图像多尺度融合方法

针对传统的热红外与可见光图像融合方法对比度低,容易出现边缘细节、目标等信息丢失或减弱的现象,提出一种顾及区域特征差异的热红外与可见光图像多尺度融合方法。首先采用自适应PCNN（脉冲耦合神经网络）模型和二维Renyi熵相结合的图像分割方法,分别对红外和可见光图像进行区域分割;然后利用非下采样Contourlet变换对原图像进行多尺度多方向分解,根据区域的特征差异设计不同的融合规则,融合热红外与可见光图像。实验结果表明,该方法不仅能有效地融合热红外图像的目标特征,还能更多地保留可见光图像丰富的背景信息,融合图像对比度高,在视觉效果和客观评价上优于传统融合方法。     

结合NSCT和KPCA的高光谱遥感图像去噪

针对高光谱遥感图像易受噪声干扰,本文提出了一种基于非下采样Contourlet变换NSCT(Nonsubsampled Contourlet Transform)和核主成分分析KPCA(Kernel Principal Component Analysis)的去噪方法。首先对高光谱各波段图像进行NSCT分解;然后利用KPCA对NSCT系数进行处理,并在KPCA重构时依据各类噪声的特性选取合适的主成分;最后用处理过的系数进行逆变换得到去噪图像。实验结果表明,本文方法抑制了高光谱遥感图像中的噪声干扰,较完整地保留了原始数据的有效信息。     

联合NSCT与多重分形的高噪声侧扫声呐图像分割

针对现有分割算法对高噪声侧扫声呐图像分割准确率低的问题,提出了一种综合利用NSCT(non-subsampled contourlet transform)分解图像、局部标准差和均值组合增强图像和多重分形判断图像奇异性的侧扫声呐图像分割方法。首先,借助NSCT分解图像,获得滤除高频噪声且保留轮廓信息的低频图像和一系列高频方向子带图像。然后,基于侧扫声呐图像中目标及其阴影伴随出现的特点,计算低频图像的局部标准差与均值的组合特征,获得分别突显目标及其阴影的特征图,使用多重分形分割方法分割特征图,获得低频图像分割结果;利用图像差分和非极大值抑制方法分割高频方向子带图像,获得高频分割结果;融合高低频分割结果获得目标及其阴影的精细边缘。最后通过试验验证了本文方法的有效性。     

面向对象的高光谱遥感影像稀疏表示分类

稀疏表示用于高光谱遥感影像分类多是基于像素层次来处理的。文中提出一种面向对象的高光谱遥感影像稀疏表示分类方法。首先从高光谱影像中提取4个波段组成标准的多波段影像,进行面向对象的影像分割;然后计算各对象在各波段上的光谱均值,并选取少量样本进行训练;最后利用基于Fisher字典学习的稀疏表示进行高光谱遥感影像的分类。实验结果表明,该方法可以利用较少的样本得到较好的分类效果,与基于像素层的稀疏分类相比较,分类精度与效率均有所提高,分类结果更接近真实地物,避免了零碎图斑。     

Indusion: Fusion of Multispectral and Panchromatic Images Using the Induction Scaling Technique

The fusion of multispectral (MS) and panchromatic (PAN) images is a useful technique for enhancing the spatial quality of low-resolution MS images. Liu recently proposed the smoothing-filter-based intensity modulation (SFIM) fusion technique. This technique upscales MS images using bicubic interpolation and introduces high-frequency information of the PAN image into the MS images. However, this fusion technique is plagued by blurred edges if the upscaled MS images are not accurately coregistered with the PAN image. In the first part of this letter, we propose the use of the Induction scaling technique instead of bicubic interpolation to obtain sharper, better correlated, and hence better coregistered upscaled images. In the second part, we propose a new fusion technique derived from induction, which is named ldquoIndusion.rdquo In this method, the high-frequency content of the PAN image is extracted using a pair of upscaling and downscaling filters. It is then added to an upscaled MS image. Finally, a comparison of SFIM (with both bicubic interpolation and induction scaling) is presented along with the fusion results obtained by IHS, discrete wavelet transform, and the proposed Indusion techniques using Quickbird satellite images.     

结合nDSM的高分辨率遥感影像深度学习分类方法

针对高分辨率遥感影像因其地物类内差异大、光谱信息相对欠缺导致现有影像分类方法存在错分现象较多、地物边界残缺不完整等问题，本文提出了一种归一化数字表面模型（nDSM）约束的高分辨率遥感影像深度学习分类方法。首先，将nDSM数据作为附加波段叠加在遥感影像上并获取训练样本；然后，利用优化的U-Net网络进行模型训练得到最优模型；最后，利用最优模型对附加了nDSM波段的遥感影像进行地物分类。试验结果表明，本文方法引入nDSM数据用于U-Net模型训练和分类，可有效提高影像分类精度，得到更加真实可靠的分类结果。     

Monitoring early stage invasion of exotic Spartina alterniflora using deep-learning super-resolution techniques based on multisource high-resolution satellite imagery: A case study in the Yellow River Delta,China

Over the past decades, Spartina alterniflora, one of the top exotic invasive plants in China, has expanded throughout coastal China. In the Yellow River Delta (YRD), the rapid expansion of S. alterniflora has caused serious negative ecological effects. Current studies have concentrated primarily on mapping the distribution of S. alterniflora with medium-resolution satellite imagery at the regional or landscape scale, which have a limited capability in early detection and monitoring of the invasive process at the patch scale. In this study, we proposed a framework for monitoring the early stage invasion of S. alterniflora patches in the YRD using multiyear multisource high-spatial-resolution satellite imagery with various ground sampling distances (WorldView-2, SPOT-6, GaoFen-1, GaoFen-2, and GaoFen-6 from 2012 to 2019). First, we proposed to use deep-learning-based image super-resolution models to enhance all images to submeter (0.5 m) resolution. Then, we adopted stepwise evolution analysis-based image segmentation and object-based classification rules to detect and delineate S. alterniflora patches from the super-resolved imagery. By investigating Super-Resolution Convolutional Neural Networks (SRCNN) and Fast Super-Resolution Convolutional Neural Networks (FSRCNN) and comparing these methods with the conventional bicubic interpolation method for image resolution enhancement, we concluded that FSRCNN was superior in constructing spectral and structural details from the 1 m/1.5 m/2 m resolution images to 0.5 m resolution. FSRCNN, in particular, was more effective and efficient in discerning and estimating the size of small S. alterniflora patches (<50 m²). Using our method, 76 of 83 field-measured small patches were accurately detected and the delineated S. alterniflora patch perimeters agreed well with the field-measured patch perimeters (root mean square error [RMSE] = 8.29 m, mean absolute percentage error [MAPE] = 23.46 %). The invasion process showed fast expansion from 2012 to 2015 and slow growth from 2016 to 2019. We observed that the landward limits of S. alterniflora patches were influenced by elevation and vicinity to tidal creeks.     

利用矢量影像法进行土地利用变化自动检测

为解决土地利用矢量图与遥感影像的变化检测问题,提出了一种基于类别的矢量图与遥感影像变化检测方法。在矢量图约束下,对遥感影像进行影像分割获取像斑;提取像斑在遥感影像上的直方图特征,采用G统计量度量像斑之间的特征距离;利用像斑与其他相同类别像斑之间的特征距离,构建单波段上像斑的类别异质度,自适应加权组合各波段上像斑的类别异质度构建像斑的类别异质度;依据最大熵方法获取各地物类别对应的异质度阈值,以类别为单位对各像斑进行变化判别,获取变化检测结果。在QuickBird遥感影像上的试验验证了本文方法的有效性,实现了矢量图与遥感影像的自动变化检测。     

Remote Sensing Image Fusion Based on Nonlinear IHS and Fast Nonsubsampled Contourlet Transform

The purpose of remote sensing image fusion is to inject the detail image extracted from the panchromatic (PAN) image into the low spatial resolution multispectral (MS) image. A novel remote sensing image fusion method based on fast nonsubsampled contourlet transform (FNSCT) and Nonlinear intensity-hue-saturation (IHS) is presented in this paper. Firstly, the Nonlinear IHS transform is performed on the multispectral image, and then the I-component representing the spatial resolution and the panchromatic image is transformed by NSCT to obtain the low frequency and high frequency. Finally, the coefficients are selected using the improved sum-modified-Laplacian (SML) method and the improved Log-Gabor filter in the low frequency and the high frequency, respectively. Experimental results show that the proposed method is the most advanced fusion method in subjective and objective evaluation, can provide more spatial information, and retain more spectral information compared with several other methods.     

混沌蜂群优化的NSST域多光谱与全色图像融合

为有效融合多光谱图像的光谱信息和全色图像的空间细节信息,提出了一种基于混沌蜂群优化和改进脉冲耦合神经网络(PCNN)的非下采样Shearlet变换(NSST)域图像融合方法。首先对多光谱图像进行Intensity-HueSaturation(IHS)变换,全色图像的直方图按照多光谱图像亮度分量的直方图进行匹配;然后分别对多光谱图像的亮度分量和新全色图像进行NSST变换,对低频分量使用改进加权融合算法进行融合,以互信息作为适应度函数,利用混沌蜂群算法找到最优加权系数。对高频分量采用改进脉冲耦合神经网络(PCNN)方法进行融合,再经NSST逆变换和IHS逆变换得到融合图像。本文方法在主观视觉效果和信息熵、光谱扭曲度等客观定量评价指标上优于基于IHS变换、基于非下采样Contourlet变换(NSCT)和非负矩阵分解(NMF)、基于NSCT和PCNN等5种融合方法。本文方法在提升图像空间分辨率的同时,有效地保留了光谱信息。     

Urban-Area Segmentation Using Visual Words

In this letter, we address the problem of urban-area extraction by using a feature-free image representation concept known as “Visual Words.” This method is based on building a “dictionary” of small patches, some of which appear mainly in urban areas. The proposed algorithm is based on a new pixel-level variant of visual words and is based on three parts: building a visual dictionary, learning urban words from labeled images, and detecting urban regions in a new image. Using normalized patches makes the method more robust to changes in illumination during acquisition time. The improved performance of the method is demonstrated on real satellite images from three different sensors: LANDSAT, SPOT, and IKONOS. To assess the robustness of our method, the learning and testing procedures were carried out on different and independent images.      

Detail focused fusion of hyperspectral and multispectral imagesEI北大核心CSCD

Hyperspectral image (HSI) and multispectral image (MSI) are two types of images widely used in the field of remote sensing. These images are useful in certain applications, such as environmental monitoring, target detection, and mineral exploration. HSI contains a large amount of spectral information. Photons are typically collected in a larger spatial area on the sensor to ensure a sufficiently high signal-to-noise ratio (SNR). Accordingly, the HSI spatial resolution is much lower compared with MSI. This low spatial resolution greatly affects the practicality of HSI. Accordingly, fusing a low-spatial resolution HSI (LR-HSI) with a high-spatial resolution MSI (HR-MSI) in the same scene to obtain a high-resolution HSI (HR-HSI) is a method for solving such problems, which resolves the contradiction that the spatial resolution and the spectral resolution cannot simultaneously maintain a high level. From the analysis of fusion effect, the spatial and spectral reconstruction errors of the existing algorithms are mainly reflected in the edge and detail areas. The method proposed in this work was a fusion algorithm for dictionary construction and image reconstruction based on detail attention. In terms of maintaining spectral characteristics, the spectral distribution in the detail area is complex and diverse because of the proximity effect of the image. This work proposes to perform dictionary learning on the image and detail layers. The detail perception error terms and a constraint of edge adaptive directional total variation are proposed for spatial characteristic enhancement, which is combined with a local low rank constraint in the same fusion framework to estimate the sparse coefficient. Experiments were conducted on two datasets, namely, Pavia University and Indian Pine, to verify the effectiveness of the proposed method. The quantitative evaluation metrics contain peak SNR, relative dimensionless global error in synthesis, spectral angle map, and universal image quality index. Based on the experimental comparison, the fusion result of the algorithm proposed in this work is significantly improved compared with those of the other algorithms in terms of spatial and spectral characteristics. This work uses dictionary learning to propose a fusion algorithm for dictionary construction and image reconstruction with attention to details through the analysis of the existing hyperspectral and multispectral image fusion algorithms. A hierarchical dictionary learning algorithm is proposed to address the problem of large reconstruction error in the detail part of the existing algorithms. The detail perception error term and the direction adaptive full variational regularization term are used to improve the spectral dictionary solution and coefficient estimation, respectively. The result of the fusion is the error in the spectral characteristics and spatial texture of the detail, which achieves an accurate representation of the edge detail. © 2022 National Remote Sensing Bulletin. All rights reserved.