基于总变分的湍流退化噪声图像盲反卷积复原

根据总变分的噪声抑制特性和大气湍流成像过程,建立了基于总变分的大气湍流噪声图像多帧盲反卷积复原最小化模型,以基于共轭梯度数值优化方法的交替迭代算法求解,复原出了观测目标的清晰图像。在计算机上模拟了湍流退化和噪声污染图像。实验结果表明,该复原算法能有效地克服大气湍流和噪声的影响,可复原出清晰的原始目标图像。     

基于噪声特性的大气湍流退化图像多帧盲反卷积复原

由于大气湍流和噪声的影响,造成观测目标图像的退化.为了目标的精确观测,根据噪声特性,结合符合物理意义的约束条件,提出了新的大气湍流图像盲反卷积复原最小化模型,并以共轭梯度数值优化方法交替迭代求解,复原观测目标图像.为验证提出的算法的有效性,在计算机上模拟参数为望远镜口径为2.0 m,大气相干长度为0.1 m,图像信噪比为10 dB的大气湍流退化和噪声污染的图像,以提出的盲反卷积复原方法复原,实验结果表明,提出的盲反卷积复原算法避免了传统的盲反卷积复原算法的缺陷,有效地克服大气湍流和噪声的影响,复原出了清晰的观测目标图像.该图像盲反卷积复原方法的研究,对地基望远镜的观测有重要的基础性作用.     

稀疏孔径光学系统成像的图像恢复算法研究

提出两种稀疏孔径光学系统成像的图像恢复模型.分析维纳滤波、最小二乘方滤波和极大似然法盲去卷积三种图像恢复算法的适用条件.针对存在噪声干扰的稀疏孔径光学系统,通过实验对比,指出维纳滤波和最小二乘方滤波把相机光学传函当作系统传函,其理论推导能够达到最优.盲去卷积把大气传输函数和相机光学传函作为系统传函进行恢复,其恢复结果优于维纳滤波带入常数K和最小二乘方滤波调整参量结果.     

改进的盲解卷积技术在远距离激光光斑图像复原中的应用

为了提高外场实验中远距离测量激光光斑位置的精度,提出利用盲解卷积技术对光斑图像进行事后复原来削弱大气湍流对光斑成像的影响。首先,介绍了经典盲解卷积算法,分析了其不足之处,并提出了一种改进的盲解卷积算法。为了提高目标函数的收敛性和收敛速度,在TV(Total Variation)目标函数加入惩罚项,并对交替迭代法进行改进。然后,用数学方法证明了改进的盲解卷积算法的收敛性。最后,进行了仿真实验。与传统算法相比,用改进算法恢复的图像信噪比至少提升了15%。文中给出了外场试验图像的实际复原效果。     

基于维纳滤波的迭代盲解卷积

盲解卷积是在两个卷积因子未知的情况下,通过卷积结果来获知卷积因子的。不考虑噪声,针对高斯模糊图像,在初始估计点扩展函数之后,利用维纳滤波的方法进行频域迭代盲解卷积,达到图像恢复的目的。实验表明,恢复的图像纹理比较清晰,边缘有所改善,主观视觉效果显著。该算法提高了分辨率。     

基于噪声方差来计算变分图像反卷积的规整化参数

采用定点迭代进行变分图像恢复并在这个计算框架下提出利用噪声方差选择规整化参数的方法.假定已知观测图像中初始噪声统计特性.为了在反卷积过程中正确地估计噪声的方差,构造一幅纯噪声图像跟实际的观测图像同步进行反卷积计算,并把纯噪声图像的方差作为观测图像中噪声方差的估计值来辅助计算规整化参数.针对规整化的各项异性,提出了能够保持两种噪声同步变化的特殊的规整化项.在能够准确知道迭代过程中图像包含噪声的方差的时候,建立了规整化参数λ与图像噪声方差之间的关系式.实验证明新的算法不但更好地抑制了噪声而且避免了过平滑,明显提高了基于定点迭代法计算变分图像恢复的适应性.     

基于GPU的多帧盲解卷积图像复原技术并行化实现

多帧盲解卷积图像复原技术能够进一步提高自适应光学图像的分辨力,但其算法比较复杂,处理耗时过长,对序列图像复原经常需要几分钟甚至几十分钟的计算时间,对实际应用造成了极大不便。为了提升算法的运行速度,改善其耗时过长的问题,通过研究和分析盲解卷积算法原理和算法结构,采用目前高速发展的中央处理器（CPU）和图形处理器（GPU）异构加速技术,主要对耗时最长的矩阵卷积运算进行优化,通过使用库函数与算法结构微调相结合的方法并行加速,实现多帧盲解卷积的图像复原算法的并行化。使用并行算法对图像进行复原处理,针对16帧以上分辨率为256256像素的空间目标图像,可以实现17的加速比,为图像复原的实时/准实时提供一种可行的方案。     

基于多通道盲识别的自适应光学图像事后处理

为实时恢复天文或空间目标的湍流退化成像,提出一种适应大气湍流动态变化的多通道自适应光学图像恢复方法.以自适应光学校正后不同时刻的目标成像作为多个通道,建立求解系统点扩散函数的线性方程,根据解出的点扩散函数利用超拉普拉斯算法,求解待观测目标的估计值.结果表明:不同时刻的点扩散函数之间存在互质关系,满足多通道盲识别的理论要求.利用建立的线性方程求解出的点扩散函数与原点扩散函数的均方误差在10^-30~10^-27量级,采用超拉普拉斯算法恢复出的目标成像与原始目标之间的均方误差在10^-5~10^-4量级.本文研究为湍流退化图像的实时恢复提供了理论基础.     

联合梯度预测与导引滤波的图像运动模糊复原

针对由相机与所摄景物之间发生相对位置移动所导致的图像运动模糊,提出了一种鲁棒的基于单幅运动模糊图像的盲反卷积算法。该方法首先通过预测图像中的较强边缘信息,实现用简单、易于求解的优化问题在傅里叶域中快速、准确地估计出点扩散函数。然后利用得到的点扩散函数,使用基于梯度约束的非盲反卷积算法复原清晰图像,同时采用一种新的边缘保持滤波器-导引滤波来消除噪声并抑制振铃效应。实验结果表明：本文的算法能够快速地从单幅运动模糊图像复原出具有清晰边缘和纹理的高质量图像,并且运算时间不超过20 s。     

基于散斑成像技术的扩展目标高分辨率复原

大气湍流对实现扩展目标的高分辨率重建具有重要影响.针对此问题,本文提出了一种改进的散斑成像算法.传统散斑成像算法在相位恢复计算中存在双谱数据量大和计算复杂等问题,改进算法利用图像的厄米特对称性和查找表技术将相位恢复和双谱计算紧密结合,通过计算截止频率内的每个空间频率点邻域双谱和添加双谱坐标约束使得双谱数据量减小.建立傅里叶频域相邻两象限共用的坐标查找表,确定双谱和相位恢复计算顺序,避免了双谱的对称操作从而使得整个计算简单易行.仿真实验结果表明:改进算法相对于双谱截切法使得双谱数据量至少减少了24%并准确恢复出目标相位谱,恢复相位谱经过傅里叶逆变换后清晰地显示了目标的轮廓和结构,再结合Labeyrie-Kroff法得到了目标的高分辨率图像;最后对实际天文图像进行处理,使恢复后图像的分辨率相对于原始图像得到明显提高,并且改进算法以更少的计算时间获得了与双谱截切法几乎同样的恢复效果.     

Compact multiframe blind deconvolution

We describe a multiframe blind deconvolution (MFBD) algorithm that uses spectral ratios (the ratio of the Fourier spectra of two data frames) to model the inherent temporal signatures encoded by the observed images. In addition, by focusing on the separation of the object spectrum and system transfer functions only at spatial frequencies where the measured signal is above the noise level, we significantly reduce the number of unknowns to be determined. This "compact" MFBD yields high-quality restorations in a much shorter time than is achieved with MFBD algorithms that do not model the temporal signatures; it may also provide higher-fidelity solutions.     

Multiframe blind deconvolution applied to diverse shift-and-add images of an astronomical object

In astronomical speckle imaging, deconvolving a shift-and-add (SAA) image has an advantage over deconvolving noisy specklegrams, because an SAA image is an integration of many specklegrams and has a relatively enhanced signal-tonoise ratio. In this paper, to reinforce the deconvolution of a single SAA image, we propose a multiframe deconvolution applied to multiple SAA images that are obtained by diversely recombining the same set of specklegrams to have different point spread functions. We have found that such diverse SAA images can be easily produced by permuting specklegrams to be processed by SAA. The results of experiments using simulated and observational data have shown a robustness of our present approach: in the previous approach of deconvolving a single SAA frame, the resulting object estimate is apt to be influenced by the given SAA frame and the estimation sometimes fails, whereas in the present approach, a reliable object image is stably reconstructed regardless of the given SAA frames.     

Restoration of atmospheric turbulence degraded images

A blind image deconvolution algorithm in the frequency domain is proposed which uses the edge-preserving method and generic bandwidth of optical system. Generic bandwidth of optical system is analyzed. With the benefits of bandwidth and edge-preserving method as compelling constraints, the algorithm cannot only suppress noise effectively but also restrict the bandwidth of point-spread function (PSF), so high-quality result can be obtained. The new algorithm is superior in handling unregistered channels. The performance of this approach is investigated with simulated data. As shown in our investigation, the algorithm can significantly alleviate the artifacts produced by the deconvolution process.     

Blind deconvolution: multiplicative iterative algorithm

A new algorithm has been developed for performing blind deconvolution on degraded images. The algorithm naturally preserves the nonnegative constraint on the iterative solutions of blind deconvolution and can produce a restored image of high resolution. Furthermore, benefiting from the multiplicative form, the algorithm is free from the instability of numerical computation. Results of applying the algorithm to simulated and real degraded images are reported.     

Single frame blind image deconvolution by non-negative sparse matrix factorization

Novel approach to single frame multichannel blind image deconvolution has been formulated recently as non-negative matrix factorization problem with sparseness constraints imposed on the unknown mixing vector that accounts for the case of non-sparse source image. Unlike most of the blind image deconvolution algorithms, the novel approach assumed no a priori knowledge about the blurring kernel and original image. Our contributions in this paper are: (i) we have formulated generalized non-negative matrix factorization approach to blind image deconvolution with sparseness constraints imposed on either unknown mixing vector or unknown source image; (ii) the criteria are established to distinguish whether unknown source image was sparse or not as well as to estimate appropriate sparseness constraint from degraded image itself, thus making the proposed approach completely unsupervised; (iii) an extensive experimental performance evaluation of the non-negative matrix factorization algorithm is presented on the images degraded by the blur caused by the photon sieve, out-of-focus blur with sparse and non-sparse images and blur caused by atmospheric turbulence. The algorithm is compared with the state-of-the-art single frame blind image deconvolution algorithms such as blind Richardson-Lucy algorithm and single frame multichannel independent component analysis based algorithm and non-blind image restoration algorithms such as multiplicative algebraic restoration technique and Van-Cittert algorithms. It has been experimentally demonstrated that proposed algorithm outperforms mentioned non-blind and blind image deconvolution methods.     

Comparative study on several blind deconvolution algorithms applied to underwater image restoration

Although the use of blind deconvolution of image restoration is a widely known concept, only few reports have discussed in detail its application to solving problem of restoration of underwater range-gated laser images. A comparative study of underwater image restoration using the Richardson-Lucy algorithm, the least-squares algorithm, and the multiplicative iterative algorithm for blind deconvolution is presented. All the deconvolution approaches use denoised underwater images and Wells’ small angle approximation theory of derived point spread function as the initial object and degradation guess, respectively. Owing the underwater no-reference imaging environment, image quality judgment based on the blur metric method is incorporated in our comparison to determine the appropriate deconvolution iteration number for each algorithm, which objectively evaluates the image restoration results. The performance of the three algorithms applied to underwater image restoration is discussed and reported.     

Non-blind image deconvolution using natural image gradient prior

Since non-blind image deconvolution is inherently ill-posed, the results of unregularized methods are often contaminated by noise and ringing artifacts. To reach a stable solution, we adopt the natural image gradient prior to regularize the latent image and obtain an improved version of the Richardson–Lucy (RL) algorithm. We use both synthetic and real world blurred images to test the proposed method. Experimental results show that the negative artifacts are significantly suppressed and the restored images are of high quality.