协作稀疏字典学习实现单幅图像超分辨率重建

字典的选择影响基于稀疏编码的图像超分辨率重建模型的重建质量。提出了一种基于协作稀疏表达的字典学习算法。在训练阶段,通过K-Means聚类算法将样本图像块划分为不同的聚类;构建基于同时稀疏约束条件的协作稀疏字典学习模型对每个聚类训练高、低分辨率字典;应用基于L_2范数的稀疏编码模型将图像超分辨率重建过程中输入图像块由低分辨率到高分辨率的映射转变为简单的线性映射,并针对不同聚类求得相应的线性映射矩阵。在重建阶段,输入图像块通过搜索与自身结构最相似的聚类来选择相应映射矩阵获得重建后的高分辨率图像。结果表明,本文算法通过改进字典学习过程实现了更好的图像超分辨率重建质量。     

稀疏阈值的超分辨率图像重建

为了解决基于字典学习的超分辨重构算法耗时过长的问题,提出了基于稀疏阈值模型的图像超分辨率重建方法。首先,将联合字典理论与图像块稀疏阈值方法相结合,训练得到高、低分辨率过完备图像字典对。接着,通过稀疏阈值OMP算法对图像特征块进行稀疏表示。然后,通过高分辨率字典重构出初始的超分辨图像。最后,通过改进迭代反投影算法对初始的超分辨图像进行全局优化,从而进一步提高图像重构质量。实验结果表明,超分辨图像重构平均峰值信噪比(PSNR)为30.1 d B,平均结构自相似度(SSIM)为0.937 9,平均计算时间为10.2 s。有效提高了超分辨重构的速度,改善了重构高分辨图像的质量。     

基于超分辨率和组稀疏表示的多聚焦图像融合

提出一种基于超分辨率结合组稀疏表示模型的多聚焦图像融合方法.首先,使用双三次插值方法增强源图像的分辨率及源多聚焦图像信息;然后采用自适应稀疏表示学习字典分别对没有明显主导方向和特定主导方向的图像块进行学习,并采用组稀疏表示模型对源多聚焦图像进行稀疏系数表示;最后采用最大l1范数来选择最终的表示系数向量.实验结果表明,所提方法克服了多聚焦图像融合易出现的低空间分辨率和模糊效果的缺点,具有更好的对比度和清晰度,主观视觉效果和客观指标均优于传统多聚焦图像融合方法,在三组图像融合结果的互信息指标上分别领先0.37、0.38和0.32.     

基于GoogLeNet和空间谱变换的高光谱图像超分辨率方法

为了提高高光谱图像的空间分辨率,提出了一种基于GoogLeNet和空间谱变换的高光谱图像超分辨率(SR)方法.设计出遥感图像的光谱SR框架,对图像中不同反射光谱进行提取;采用GoogLeNet的稀疏编码对粗像素光谱进行放大,并投影到高分辨率字典上,将潜在SR表示进行反转,以获得超分辨光谱;为了提高图像重构的保真度,利用...     

结合稀疏编码和空间约束的红外图像聚类分割研究

提出了结合稀疏编码和空间约束的红外图像聚类分割新算法, 在稀疏编码的基础上融合聚类算法, 扩展了传统的基于K-means聚类的图像分割方法. 结合稀疏编码的聚类分割算法能有效融合图像的局部信息, 便于利用像素之间的内在相关性, 但是对于分割会出现过分割和像素难以归类的问题.为此, 在字典的学习过程中, 将原子的聚类算法引入其中, 有助于缩减字典中原子所属类别的数目, 防止出现过分割; 考虑到像素及其邻域像素具有类别属性一致性的特点, 引入了空间类别属性约束信息, 并给出了一种交替优化算法. 联合学习字典、稀疏系数、聚类中心和隶属度, 将稀疏编码系数同原子对聚类中心的隶属程度相结合, 构造像素归属度来判断像素所属的类别. 实验结果表明, 该方法能够有效提高红外图像重要区域的分割效果, 具有较好的鲁棒性. 关键词： 图像分割 稀疏编码 聚类 空间约束     

基于双字典和稀疏表示的医学图像超分辨率重建

在医学影像图像处理过程中,由于成像技术和成像时间的限制,还无法获取满足诊断需求的清晰图像,这使得在现有技术和极短时间内所获取的医学病理图像需要进行超分辨率的重建处理;基于学习的图像超分辨率思想是从已建立的先验模型中重建出高频细节;在文章中,将要估计的高频信息认为是由主要高频和冗余高频两部分组成,提出了一种基于双字典学习和稀疏表示的医学图像超分辨率重建算法,由主要字典学习和冗余字典学习组成,分别渐近地恢复出主要高频细节和冗余高频细节;实验结果的数据分析和视觉效果显示,所提出双层递进方法能够恢复更多的图像细节且在性能指标上比现有的其他几种方法均有所提高。     

基于稀疏表示和近邻嵌入的图像超分辨率重构

提出基于稀疏表示和近邻嵌入的单帧图像超分辨率重构算法。为低分辨率和高分辨率图像块训练两个基于稀疏表示的过完备字典,在训练的低分辨率图像块和高分辨率图像块中分别选取与这两个字典原子最近的图像块近邻,通过图像块近邻来计算构图像块的权重。一旦得到权重矩阵,高分辨率重构图像块可以由低分辨率图像块与相应权重相乘来表示。与之前的算法相比,所提出的算法在计算字典原子与图像块距离的时候不是逐个图像块进行计算,而是先将图像块聚类,计算字典原子与类中心的距离,在距离最近的一类中选取图像块。计算权重矩阵的时间可以大大减少,提高计算效率。所得到的PSNR与其它算法相比,也有一定提高。     

基于判别字典学习与形态成分分解的多源图像融合

提出了一种基于形态成分分析的多源图像融合方法。为了将源图像中不同形态结构的卡通-纹理成分分离,把图像的分解问题转化为图像的分类问题,设计了卡通纹理判别字典学习模型。考虑到图像分解不仅与字典有关,还与分解的策略有关,设计了一种新的图像分解模型。在模型中,将纹理成分看成叠加在源图像卡通成分上的噪声,引入非局部均值相似性的一致性正则项,来约束稀疏编码系数的解空间。根据对应成分的编码系数l1范数值最大来选取融合图像的编码系数。实验结果表明,无论是在视觉效果还是在客观指标上,方法都具有更好的融合性能。     

凝视成像降质模型的超分辨率重建

针对吉林一号视频03星凝视成像的特点,为提高其图像空间分辨率,解决光学系统分辨率不足的问题,提出一种视频卫星超分辨率重建的新算法——凸集中间映射。以主动观点分析视频卫星凝视成像特点,建立了基于凝视成像的图像降质模型。为求解该降质模型的逆过程,以凸集理论为基础,建立了基于中间降质过程的约束集和相应的点投影算子,通过点投影算子逐帧修正高分辨率图像灰度值,最终将重建的高分辨率图像约束于凸集的交集上。实验结果表明,该算法使图像分辨率提高近30%,克服了同类算法具有投影误差和重叠伪影的缺点,图像质量评价指标均优于所列其他算法,8帧重建得到收敛解,对不同清晰度的图像重建均具有可行性和稳健性。说明该算法适用于视频卫星图像超分辨率重建。     

基于多相组重建的航空图像超分辨率算法

为提高航空图像的空间分辨率, 提出一种基于多相组重建的超分辨率算法. 融合图像间的互补信息, 将多帧低分辨率图像作为图像基, 参考帧分解为多相组, 利用差异采样特性构建图像基与参考帧之间的的多相组线性关系重建得到高分辨率图像的多项组, 经图像多相分解逆变换获得融合的高分辨率图像. 根据该融合图像的局部内容和结构信息自适应调整控制核核函数, 应用改进的控制核回归算法去除图像模糊和噪声得到清晰的超分辨率图像. 与传统算法相比, 该算法无需图像配准和迭代过程, 计算效率极大地提高. 实验结果表明, 本文算法能够有效提高航空图像的空间分辨率, 在定量评价指标和主观视觉效果方面都有显著提高.     

A sparse Bayesian representation for super-resolution of cardiac MR images

High-quality cardiac magnetic resonance (CMR) images can be hardly obtained when intrinsic noise sources are present, namely heart and breathing movements. Yet heart images may be acquired in real time, the image quality is really limited and most sequences use ECG gating to capture images at each stage of the cardiac cycle during several heart beats. This paper presents a novel super-resolution algorithm that improves the cardiac image quality using a sparse Bayesian approach. The high-resolution version of the cardiac image is constructed by combining the information of the low-resolution series –observations from different non-orthogonal series composed of anisotropic voxels – with a prior distribution of the high-resolution local coefficients that enforces sparsity. In addition, a global prior, extracted from the observed data, regularizes the solution. Quantitative and qualitative validations were performed in synthetic and real images w.r.t to a baseline, showing an average increment between 2.8 and 3.2 dB in the Peak Signal-to-Noise Ratio (PSNR), between 1.8% and 2.6% in the Structural Similarity Index (SSIM) and 2.% to 4% in quality assessment (IL-NIQE). The obtained results demonstrated that the proposed method is able to accurately reconstruct a cardiac image, recovering the original shape with less artifacts and low noise.     

The infrared and visible image fusion algorithm based on target separation and sparse representation

Although the fused image of the infrared and visible image takes advantage of their complementary, the artifact of infrared targets and vague edges seriously interfere the fusion effect. To solve these problems, a fusion method based on infrared target extraction and sparse representation is proposed. Firstly, the infrared target is detected and separated from the background rely on the regional statistical properties. Secondly, DENCLUE (the kernel density estimation clustering method) is used to classify the source images into the target region and the background region, and the infrared target region is accurately located in the infrared image. Then the background regions of the source images are trained by Kernel Singular Value Decomposition (KSVD) dictionary to get their sparse representation, the details information is retained and the background noise is suppressed. Finally, fusion rules are built to select the fusion coefficients of two regions and coefficients are reconstructed to get the fused image. The fused image based on the proposed method not only contains a clear outline of the infrared target, but also has rich detail information.     

合成孔径压缩感知超声成像中的高效能稀疏字典设计

针对目前常见的稀疏字典缺乏针对性,在合成孔径医学超声成像中的应用效果不佳,难以在低压缩率下保证重构图像质量的问题,本文设计了一种高效能的稀疏字典。根据超声回波信号是由发射脉冲信号经过不同延时衰减后叠加的特点,利用发射脉冲作为基函数构造稀疏字典,回波信号在该稀疏字典确定的变换域中具备很好的稀疏性,理论上能使其稀疏表示系数的稀疏度等于超声阵元接收到的反射回波数。通过FieldⅡ对简单点目标和复杂目标的仿真结果表明:在相同的重构算法和压缩率下该稀疏字典重构的平均绝对误差明显小于常见的稀疏字典,其值仅为DWT的几分之一,DFT和DCT的几十分之一,能让回波信号以更低的压缩率实现相同的恢复效果。本文最后使用体模的实际采集数据对算法的实际效果进行检测,实验结果也与仿真结果基本一致。基于该稀疏字典的压缩感知算法可以进一步减少合成孔径成像所需存储的数据量、降低系统的复杂度。      

Visual attention guided image fusion with sparse representation

Image fusion techniques aim at transferring useful information from the input source images to the fused image. The common assumption for most fusion approaches is that the useful information is defined by local features such as contrast, variance, and gradient. However, there is no consideration of global visual attention of the whole source images which indicates the “interesting” information of the source images. In this paper, we firstly review the patch-based image fusion methods which attract the attention and interest of many researchers. Then, a visual attention guided patch-based image fusion method is proposed. The visual attention maps of the source images are calculated from the sparse represent coefficients of the source images. Then, the sparse coefficients are fused with the guidance of visual attention maps in order to emphasize the global “interesting” objects in the source images. Finally, the fused image is reconstructed from the fused sparse coefficients. The new fusion strategy ensures that the objects being “interesting” for our visual system are preserved in the fused image. The proposed approach is tested on infrared and visual, medical, and multi-focus images. The results compared with those of traditional methods show obvious improvement in objective and subjective quality measurements.     

基于稀疏表示的低比特率可伸缩图像编码算法研究

根据图像的几何结构特性,从人类视觉系统特性出发,建立了Gabor感知多成份字典,进而模拟人类视觉通路的层次处理机制,构建了稀疏编码网络,能够有效去除图像中的高阶冗余,形成更为稀疏的表示。对稀疏表示系数重组后进行比特平面量化,实现了低比特率的可伸缩编码。实验结果表明,在低比特率下,本文算法压缩后重构图像的感知质量要明显优于JPEG2000,峰值信噪比也与其相当,并且对于图像中的边缘和纹理等细节保持效果更佳。     

基于空时联合稀疏重构的红外小弱运动目标检测方法

提出了一种基于空时联合稀疏重构的红外小弱运动目标检测算法。通过学习序列图像内容而构建的空时联合字典能同时刻画目标或背景的形态特征和运动信息;利用多元高斯运动模式从空时联合字典中提取出目标空时字典和背景空时字典,目标空时过完备字典描述移动的目标,背景空时过完备字典表征背景噪声。将连续多帧图像在空时联合字典上进行稀疏分解,然后分别利用目标空时字典和背景空时字典中的最大稀疏系数及其空时原子重构信号,获取重构残余能量差异来区分目标和背景。试验结果表明,由同源的空时字典重构的残余能量小,而由异构的空时字典恢复的残余能量大,该方法不仅能提高序列信号表示的稀疏度,还能有效提高小运动目标的探测能力。     

Convex relaxation based sparse algorithm for hyperspectral target detection

Target detection in hyperspectral images is an important task. In this paper, we propose a sparsity based algorithm for target detection in hyperspectral images. In sparsity model, each hyperspectral pixel is represented by a linear combination of a few samples from an overcomplete dictionary, and the weighted vector for such reconstruction is sparse. This model has been applied in hyperspectral target detection and solved with several greedy algorithms. As conventional greedy algorithms may be trapped into a local optimum, we consider an alternative way to regularize the model and find a more accurate solution to the model. The proposed method is based on convex relaxation technique. The original sparse representation problem is regularized with a properly designed weighted ?₁ minimization and effectively solved with existing solver. The experiments on synthetic and real hyperspectral data suggest that the proposed algorithm outperforms the classical sparsity-based detection algorithms, such as Simultaneous Orthogonal Matching Pursuit (SOMP) and Simultaneous Subspace Pursuit (SSP) and conventional ?₁ minimization.     

Manifold-preserving single-image super-resolution based on collaborative representation support

Nowadays, super-resolution is becoming more and more important in most optical imaging systems and image processing applications due to current resolution limit of charged couple device (CCD) and complementary metal-oxide semiconductor (CMOS). In this paper, we proposed a novel single-image super-resolution algorithm, which combines collaborative representation into manifold-preserving approach. The main contributions of our work can be summarized into two points. First, supporting bases which are used to calculate the mapping relationship between low-resolution (LR) images and high-resolution (HR) images are obtained by applying collaborative representation on the neighborhoods of the dictionary atoms, where the neighborhoods are clustered from the whole training sample pools. Second, to achieve a balance between execution speed and reconstruction quality, a global solution for our framework is constructed by transferring the online calculating process to offline. We demonstrate better results on commonly used datasets, showing both better visual performance and higher index values compared to other methods.