自适应加权的总变分正则化图像超分辨率重建

在Farsiu提出的双边总变分正则化方法中,尺度加权系数和正则化参数为定值,在边缘、纹理区域,重建图像效果不理想。针对这个问题,提出了一种自适应加权正则化函数和正则化参数的重建算法,该方法利用图像局部结构信息控制权函数形状、带宽和正则化参数,使这些参数根据图像局部结构信息自适应地调整。对所提出的算法进行了仿真实验,实验结果表明,与传统的总变分重建方法相比较,该算法能更好地重建图像的纹理细节,重建图像的对比度高。     

基于二阶广义全变差的多帧图像超分辨率重建

图像超分辨率重建是图像处理领域的重要问题.本文将二阶广义全变差用于基于正则化的多帧图像超分辨率重建问题,构建了基于二阶广义全变差正则项的图像超分辨率模型.为了更好地保持重建图像的边缘和细节,采用图像空域自适应正则化参数,并针对该重建模型的非光滑性,给出了基于半二次正则化和交替方向法的求解算法.实验结果表明该模型和数值算法能够较好地提高图像的分辨率,同时可以较好地保持图像的细节信息.     

基于Otsu阈值分割的边缘快速图像插值算法

为了满足视频监控、目标检测与识别过程中较高图像质量和较低算法复杂度要求,以及改善传统图像插值中细节模糊和边缘锯齿效应,文中提出一种基于Otsu阈值分割的边缘快速图像插值算法。利用Otsu算法,根据目标区域和背景区域的类方差最大,确定分割阈值,对非边缘区域进行双线性插值,边缘区域利用与待插值点周围6个或8个相邻降采样像素局部结构的多方向特点,自适应估计高分辨率像素值。实验表明,该算法运算复杂度低,很好保持了图像的边缘,获得了视觉质量较好的高分辨率图像。     

自适应约束下的双边全变差正则化超分辨率重建

在经典的双边全变差( BTV)超分辨率重建中,加权系数和正则化参数的恒定性导致重建结果边缘保持能力受限。为此,提出了一种自适应约束的BTV正则化先验模型。算法首先定义了图像的局部邻域残差均值以区分当前像素属于平坦区域还是边缘区域;然后针对加权系数的不变性导致边缘削弱的问题,利用边缘方向和垂直边缘方向扩散性的不同,设计自适应权重矩阵;最后根据代价函数的极值问题推导出迭代公式,从而进行图像的超分辨率重建,重建过程中采用自适应的方法确定正则化参数,以便求得代价函数的全局最优解,提高了算法的鲁棒性。实验结果表明：与双三次线性插值法和经典BTV算法相比,该算法取得了更好的视觉效果和更高的峰值信噪比,更多地保留了图像的边缘细节信息。     

烟花算法在SAR图像属性散射中心参数估计中的应用

针对合成孔径雷达（SAR）属性散射中心估计问题,提出基于烟花算法的方法。首先,在图像域对SAR图像中高能量区域进行分割解耦,获得单个独立散射中心在图像域的表现形式。在此基础上,以属性散射中心参数化模型为基础,构建优化问题,对分离出来的单个散射中心进行最优参数的搜索。在此阶段,引入烟花算法进行参数寻优。该算法具有强大的全局和局部搜索能力,在保证优化精度的条件下避免陷入局部最优,从而保证散射中心参数估计的可靠性。在原始图像中剔除求解后的单个散射中心,对残余图像进行高能量区域分割,序惯估计下一个散射中心的属性参数。最终,获取输入SAR图像上所有散射中心的参数集。实验中,首先基于MSTAR数据集中的SAR图像进行参数估计验证,通过参数估计结果与原始图像的对比以及基于估计参数集对原始图像进行重构,反映了提出算法的有效性。此外,实验还基于估计得到的属性参数进行SAR目标识别算法验证,通过与其他参数估计算法在相同条件下进行识别性能的对比,进一步体现了提出方法在属性散射中心参数估计上的性能优势。     

双层约束下基于局部和全局信息的图像插值新模型

该文提出一种双层约束的图像插值模型,模型在原始未插值图像梯度模约束下同时基于局部和全局信息处理。使用偏微分方程处理边缘像素,锐化边缘同时平滑边缘块状效应;平滑区域像素点的插值操作使用非局部均值模型,非局部均值模型通过对原始图像全局信息加权平均得到待处理图像像素值,图像平滑。使用双层约束模型处理纹理图像可以保持纹理特征,平滑纹理部分线形特征位置的块状效应。最后理论和实验结果证明使用双层控制模型可以直接将噪声图像插值放大。     

基于结构保持MRF模型的SAR图像去斑

该文提出一种用于SAR图像贝叶斯去斑的,具有结构保持特性的马尔可夫随机场模型。该模型通过引入一组权重参数建模待处理像素与其邻域点的空间相关性,可有效描述图像中的各种局部特性。准确的参数估计是模型有效运用的关键,借鉴随机松弛算法的思想,通过引入温度变量,实现了迭代过程中权重参数的自适应估计。仿真和实际SAR图像数据的实验结果,验证了所提模型及参数估计方法的有效性。     

基于EM算法的SAI图像插值改进技术

提出了一种基于期望最大（Expectation Maximization,EM）算法对自适应二维自回归建模与软判决估计联合插值（SAI）算法进行改进的算法,它将插入像素的估计值反馈回分段二维自回归模型（PAR）参数的估计过程中,修正估计出的模型参数。实验结果显示：该算法能产生良好的视觉效果及较高的峰值信噪比（PSNR）。     

基于局部非局部联合模型的压缩图像插值技术

本文提出了一种基于局部自回归模型和非局部自相似模型的正则化的压缩图像插值技术.传统的基于图像先验模型正则化图像插值技术存在着2个缺陷.一方面,通常是只利用一个图像的先验特性,不能得到视觉质量很好的超分辨率效果;另一方面,在描述图像的非局部自相似特性时,多数利用一种相似块加权的方式来描述当前块,没能够将具有相同纹理的一系列的相似块的特性描述完整.基于以上2点考虑,研究设计整合了2种不同的模型:局部自回归模型和非局部自相似模型,形成一个整体的正则化的框架.不同于传统的只利用高低分辨率之间几何二元性的自回归模型,本文提出了一种自适应加权的在高分辨率图像上迭代的自回归模型;而非局部的自相似模型,并且以相似块组成的一个三维数据结构的变换域稀疏性来对一系列的相似块统一描述.由于压缩图像的特点,研究针对压缩图像提出了软数据精度项,最终采用分离布莱格曼方法来求解整体的正则化目标函数.     

基于核回归正则项的图像超分辨率重建

讨论了图像成像的基本模型,并提出了一种基于调整核回归函数作为正则项的序列图像重建算法。该算法是对已经提出的核回归算法的改进,减少其在超分辨率图像重建时的运算量。而且在图像配准过程中针对图像间只存在平移和旋转变换,采用了基于矩形像素值的亚像素配准方法,以提高配准的速度和精度。利用此算法对序列图像进行重建仿真,并通过结论得出其在噪声严重的情况下具有更好的边缘保留特性。     

融入结构信息的稀疏低秩丰度估计算法研究

丰度估计(AE)是从高光谱图像中识别地物的关键预处理技术.鉴于线性混合模型的可解释性以及数学上的可操作性,带约束的线性回归技术在丰度矩阵估计中备受关注.目前,这类方法存在的缺陷是其拟合过程中仅仅考虑到估计数据与真实数据之间的拟合误差,忽略了估计数据的结构与真实数据的结构之间的相似性信息.因此,提出了融合结构信息的线性回归模型,并应用于稀疏低秩丰度矩阵估计领域.首先,通过增加结构信息改进传统的带约束的线性回归模型,并经数学理论证明了增加结构信息的模型较传统模型更加有效;其次,应用该方法改进稀疏低秩丰度估计的数学模型;最后,采用交替乘子法(ADMM)技术求解新模型.实验结果表明,融入结构信息的稀疏低秩丰度估计算法能够有效地提高仿真数据和实际高光谱数据的丰度估计的估计精度,改善其抗噪性能.     

MSE-based regularization approach to direction estimation ofcoherent narrowband signals using linear prediction

This paper addresses the problem of directions of arrival (DOAs) estimation of coherent narrowband signals impinging on a uniform linear array (ULA) when the number of signals is unknown. By using an overdetermined linear prediction (LP) model with a subarray scheme, the DOAs of coherent signals can be estimated from the zeros of the corresponding prediction polynomial. Although the corrected least squares (CLS) technique can be used to improve the accuracy of the LP parameters estimated from the noisy array data, the inversion of the resulting matrix in the CLS estimation is ill-conditioned, and then, the CLS estimation becomes unstable. To combat this numerical instability, we introduce multiple regularization parameters into the CLS estimation and show that determining the number of coherent signals is closely related to the truncation of the eigenvalues. An analytical expression of the mean square error (MSE) of the estimated LP parameters is derived, and it is clarified that the number of signals can be determined by comparing the optimal regularization parameters with the corresponding eigenvalues. An iterative regularization algorithm is developed for estimating directions without any a priori knowledge, where the number of coherent signals and the noise variance are estimated from the noise-corrupted received data simultaneously     

面向高分辨SAR成像的正则参数自学习

针对基于正则优化的高分辨SAR成像惩罚项系数自学习难点问题,本文提出一种贝叶斯边缘估计（Marginal Estimation Bayes,MEB）算法,以实现目标多先验模型的高精度特征拟合,提升成像特征恢复精度。该方法根据观测数据进行交替方向乘子法（Alternating Direction Method of Multipliers,ADMM）凸优化框架建模,并利用贝叶斯理论推导参数的最大边缘似然分布,同时采用Moreau-Yoshida未经调整的朗之万算法（Moreau-Yoshida Unadjusted Langevin Algorithm,MYULA）实现后验采样求解,引入梯度投影法解决正则参数自学习问题,最后利用自学习参数进行优化成像。该算法可实现多正则项优化多参数协同自适应参数估计。另外,针对可能存在的目标先验非可微问题,本文利用近端算法中的次梯度优化,通过邻近算子来求解非可微先验的次梯度,可实现非可微正则函数的参数自学习。实验部分利用点目标仿真与美国Sandia实验室公布的实际数据。实验结果表明,相对于遍历最优结果,本文所提方法得到的结果与最优值的误差均在15%之内。另外,通过相变热力图（Phase Transition Diagram,PTD）定量验证了算法的有效性,同时将本文算法与其他自学习算法进行对比,验证了算法的实用性。      

Data fusion: color edge detection and surface reconstructionthrough regularization

Data fusion provides tools for solving problems which are characterized by distributed and diverse information sources. In this paper, the authors focus on the problem of extracting features such as image discontinuities from both synthetic and real images. Since edge detection and surface reconstruction are ill-posed problems in the sense of Hadamard, Tikhonov's regularization paradigm is proposed as the basic tool for solving this inversion problem and restoring well-posedness. The proposed framework includes: (1) a systematic view of oneand two-dimensional regularization; (2) extension of the standard Tikhonov regularization method by allowing space-variant regularization parameters; and (3) further extension of the regularization paradigm by adding multiple data sources to allow for data fusion. The theoretical approach is complemented by developing a series of algorithms, then solving the early vision problems of color edge detection and surface reconstruction. An evaluation of these methods reveals that this new analytical data fusion technique output is consistently better than each of the individual RGB edge maps, and noisy corrupted images are reconstructed into smooth noiseless surfaces     

A stochastic deconvolution method to reconstruct insulin secretionrate after a glucose stimulus

Insulin secretion rate (ISR) is not directly measurable in man but it can be reconstructed from C-peptide (CP) concentration measurements by solving an input estimation problem by deconvolution. The major difficulties posed by the estimation of ISR after a glucose stimulus, e.g., during an intravenous glucose tolerance test (IVGTT), are the ill-conditioning of the problem, the nonstationary pattern of the secretion rate, and the nonuniform/infrequent sampling schedule. In this work, a nonparametric method based on the classic Phillips-Tikhonov regularization approach is presented. The problem of nonuniform/infrequent sampling is addressed by a novel formulation of the regularization method which allows the estimation of quasi time-continuous input profiles. The input estimation problem is stated into a Bayesian context, where the a priori known nonstationary characteristics of ISR after the glucose stimulus are described by a stochastic model. Deconvolution is tackled by linear minimum variance estimation, thus allowing the derivation of new statistically based regularization criteria. Finally, a Monte-Carlo strategy is implemented to assess the uncertainty of the estimated ISR arising from CP measurement error and impulse response parameters uncertainty     

Direct reconstruction of kinetic parameter images from dynamic PET data

Our goal in this paper is the estimation of kinetic model parameters for each voxel corresponding to a dense three-dimensional (3-D) positron emission tomography (PET) image. Typically, the activity images are first reconstructed from PET sinogram frames at each measurement time, and then the kinetic parameters are estimated by fitting a model to the reconstructed time-activity response of each voxel. However, this "indirect" approach to kinetic parameter estimation tends to reduce signal-to-noise ratio (SNR) because of the requirement that the sinogram data be divided into individual time frames. In 1985, Carson and Lange proposed, but did not implement, a method based on the expectation-maximization (EM) algorithm for direct parametric reconstruction. The approach is "direct" because it estimates the optimal kinetic parameters directly from the sinogram data, without an intermediate reconstruction step. However, direct voxel-wise parametric reconstruction remained a challenge due to the unsolved complexities of inversion and spatial regularization. In this paper, we demonstrate and evaluate a new and efficient method for direct voxel-wise reconstruction of kinetic parameter images using all frames of the PET data. The direct parametric image reconstruction is formulated in a Bayesian framework, and uses the parametric iterative coordinate descent (PICD) algorithm to solve the resulting optimization problem. The PICD algorithm is computationally efficient and is implemented with spatial regularization in the domain of the physiologically relevant parameters. Our experimental simulations of a rat head imaged in a working small animal scanner indicate that direct parametric reconstruction can substantially reduce root-mean-squared error (RMSE) in the estimation of kinetic parameters, as compared to indirect methods, without appreciably increasing computation.     

单目图像序列中基于正则化的局部三维非刚体运动估计

通过局部的三维非刚体运动估计进而达到全局估计结果是三维非刚体运动估计中的重要方法。本文提出了在单目图像序列中利用正则化的手段解决局部的三维非刚体运动估计。首先，在帧间特征点匹配已确立的前提下，利用仿射运动模型并结合中心投影方式提出了运动估计的最小二乘模型；然后针对三维运动估计的不适定性提出了正则化的运动估计方法，以正则化的形式融入运动的先验知识，使运动估计的结果更具鲁棒性；最后利用Levenberg-Marquart方法实现运动参数的求解。仿真图像序列的实验反映了本文方法的有效性。     

基于SVM模型的单目红外图像深度估计

提出一种通过非线性学习模型来估计单目红外图像深度的算法。该算法首先通过逐步线性回归和独立成分分析(ICA)寻找对于红外图像深度相关性较强的特征,然后以具有核函数的非线性支持向量机(SVM)为模型基础,采用监督学习的方法对红外图像深度特征进行回归分析并训练,在训练过程中通过已知数据回归后的最小均方误差对模型参数进行修正,训练后的模型可对单目红外图像的深度分布进行估计。实验结果证明,利用该模型能较一致地估计单目红外图像的深度信息。     

Regression on the basis of nonstationary Gaussian processes with Bayesian regularization

We consider the regression problem, i.e. prediction of a real valued function. A Gaussian process prior is imposed on the function, and is combined with the training data to obtain predictions for new points. We introduce a Bayesian regularization on parameters of a covariance function of the process, which increases quality of approximation and robustness of the estimation. Also an approach to modeling nonstationary covariance function of a Gaussian process on basis of linear expansion in parametric functional dictionary is proposed. Introducing such a covariance function allows to model functions, which have non-homogeneous behaviour. Combining above features with careful optimization of covariance function parameters results in unified approach, which can be easily implemented and applied. The resulting algorithm is an out of the box solution to regression problems, with no need to tune parameters manually. The effectiveness of the method is demonstrated on various datasets.     

Iterative Decomposed OFDM Channel Estimation Algorithm Over Highly Mobile Channels

In orthogonal frequency-division multiplexing systems, the temporal channel gains to estimate are much more than the observable data over highly mobile channels. The basis expansion model (BEM) has been employed to reduce the number of these channel parameters. In the absence of channel statistics, generalized complex-exponential BEM (GCE-BEM) is popular for its fast algebra operation and easy generation of basis matrix. However, there is still much potential for performance improvement by modeling error reduction. In this paper, the factors affecting the modeling error are analyzed and an iterative decomposed estimation algorithm is proposed to improve the modeling accuracy. The proposed algorithm decomposes each tap into the linear part and the non-linear part. The linear part with two parameters (the middle value and the slope) is initialized by estimation in linearly time-varying channel models. And the non-linear part is addressed by the conventional least-squares (LS) method based on GCE-BEM and then the slopes of the linear part are updated for the next iteration by two distinct slope update methods. The simulations show that the proposed algorithm outperforms the conventional estimation methods with significantly reduced modeling error under both high signal to noise ratio and Doppler shift conditions.