Image segmentation via image decomposition and fuzzy region competition

Taking into account the morphological diversity of images, this paper presents a novel multiphase image segmentation method that combines image decomposition and fuzzy region competition into a unified model. To efficiently solve the minimization of the energy functional, we design an optimal iteration algorithm which integrates a modified cartoon-texture dictionary learning algorithm and wavelet shrinkage. Compared with the classical fuzzy region competition method, the proposed method not only improves the overall segmentation results, but also has more strong robustness. A series of experimental results demonstrate the applicability and effectiveness of the proposed method.     

一种更一般全变分图像复原模型

分析了TV模型图像复原的特点,针对Aubert TV模型扩散函数不满足次优条件的缺点,对扩散函数进行改进以获得更好的图像复原效果。实验结果表明,本模型在图像复原中既能克服平滑区域的阶梯效益,又能较好地保留图像的细节,取得较高的峰值信噪比,取得了很好的图像复原性能。     

基于噪声与图像同步迭代来确定时间步进法的规整化参数

为了在反卷积过程中正确地估计噪声的方差,该文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,该文提出了能够保持两种噪声同步变化的特殊的规整化项。新的规整化项在迭代纯粹噪声图像时使用,这样确保每次迭代都可以保持人工噪声与实际图像噪声的统计特性相一致。在能够准确知道迭代过程中图像包含噪声的方差的时候,该文建立了规整化参数与图像噪声方差之间的关系式并转化成简单的解一元二次方程问题。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,基于时间步进法计算变分图像恢复的适应性被明显的提高了。     

基于各向异性规整化的总变分盲复原算法研究

针对大气湍流退化图像复原问题,提出了一种基于各向异性和非线性规整化的总变分盲复原新算法,该算法主要结合图像和湍流点扩展函数的一些性质采用基于各向异性的空间自适应规整化处理,建立了具有非线性和空间各向异性的规整化函数,使其在恢复目标图像和估计点扩展函数时能自适应地进行梯度平滑。最后,通过交替最小化方案来极小化代价函数和通过定点迭代策略将非线性方程进行线性化处理,快速地估计点扩展函数和恢复图像。在微机上对数字模拟和实际退化图像进行了一系列恢复实验,验证了算法的有效性和稳健性。     

基于双参数正则化方法的图像复原

针对高斯模糊图像的纹理和结构特性,提出一种基于总变分的双参数正则化约束的图像复原方法.首先采用基于各向异性的梯度处理方法,建立了具有点扩散函数和图像像素的灰度梯度.然后进行非线性和空间各向异性的规整化处理,使其在恢复目标图像和估计点扩展函数中能自适应地进行梯度运算.最后采用交替迭代的数值处理方法提高图像复原的速度.实验结果表明,基于梯度的双参数正则化图像复原能较好地提高图像复原的效果.     

Compressive sensing reconstruction via decomposition

When recovering images from a small number of Compressive Sensing (CS) measurements, a problem arises whereby image features (e.g., smoothness, edges, textures) cannot be preserved well in reconstruction, especially textures at small-scale. Since the missing information still remains in the residual measurement, we propose a novel Decomposition-based CS-recovery framework (DCR) which utilizes residual reconstruction and state-of-the-art filters. The proposed method iteratively refines residual measurement which is closely related to the denoise-boosting techniques. DCR is further incorporated with a weighted total variation and nonlocal structures in the gradient domain as priors to form the proposed Decomposition based Texture preserving Reconstruction (DETER). We subsequently demonstrate robustness of the proposed framework to noise and its superiority over the other state-of-the-art methods, especially at low subrates. Its fast implementation based on the split Bregman technique is also presented.     

图像分解的多尺度变分模型

该文提出了一种新的多尺度变分图像分解模型。首先在Tadmar的分层多尺度变分模型的基础上,给出了一种新的(BV, H-1)分层多尺度图像分解方法,然后在逆尺度空间上积分尺度图像并将拉普拉斯算子作用于曲率项就得到了新的积分微分方程。该方程包含一个单调递增的尺度函数,它的值与残差图像的星范数成反比。接着讨论了该方程的重要性质,并给出了数值离散算法。理论分析与数值实验表明新的积分微分方程是一种有效的图像分解模型。     

一种全变差正则化流场的图像抑噪方法

利用Meyer的图像分解理论,提出一种磨光流场的全变差正则化抑噪方法。该方法首先引入负指数Hilbert- Sobolev范数度量逼近项,对图像水平曲线的法向量场进行全变差正则化磨光,然后构造出一个曲面拟合模型,拟合磨光后的流场。最后,利用有限差分法对各模型所导出Euler-Lagrange方程进行数值求解。实验结果表明,该方法在有效去噪的同时,使边缘和纹理信息均得到较好的保持。     

基于变分法的图像分割的数值算法

解释一种图像分割问题的数学模型，并讨论图像分割问题与Mumford-Shah泛函的关系，对此问题首先就一维的情形提出一种可行的解法并推广到二维。若干试验结果验证了算法的有效性。     

基于奇异值分解的图像匹配方法

传统的图像匹配方法中, 由于实时图和参考图之间存在着灰度差异和几何形变,仅用灰度作为特征进行匹配算法的性能很容易受到影响。文中提出了一种基于奇异值分解的图像匹配方法。该方法首先利用奇异值分解方法,求出模板图像矩阵的奇异值及奇异值向量,用它们作为模板图像的特征代替传统算法中的灰度对两幅待匹配图像进行全局搜索定位。由于奇异分解方法所特有的优越性,匹配实验取得了良好效果。实验结果验证了该方法的有效性。     

基于修正小波阈值的图像变分分解

Daubechies等人(2004)首先提出了图像的变分分解和小波软阈值之间的联系。小波软阈值会对图像边缘造成过度光滑,使重构图像在边缘附近产生吉布斯震荡现象,为克服该问题,本文用具有更高正则性的分段n次多项式小波阈值和指数阈值做图像分解,得到图像分解的变分泛函的近似最小值。当n越大时,图像分解的变分问题的近似最小值越逼近精确最小值。这样得到了图像的变分分解和修正小波阈值之间的联系。实验结果表明该模型用于图像分解的有效性。     

基于非局部全变差模型的图像复原方法

面向图像复原问题,本文提出一种非局部的全变差图像复原方法。该方法将传统的全变差模型拓展为非局部变差模型,充分利用非局部子块对结构的保持作用,进一步提高复原图像的质量。此外,为了解决上述的非局部全变差模型,本文引入算子运算简化目标函数, 再利用迭代的Splitting算法对其进行交替求解,提高收敛精度。实验结果表明,本文算法在视觉效果和客观评价指标两方面均优于传统算法。     

利用Talbot效应实现图象分解

Ｔａｌｂｏｔ效应可以用于周期物成象而不需要任何成象元件，并可以用来实现图象分解，本文给出了图像分解的理论分析和实验结果。     

Image blind restoration based on degradation representation network

Most deep learning (DL)-based image restoration methods have exploited excellent performance by learning a non-linear mapping function from low quality images to high quality images. However, two major problems restrict the development of the image restoration methods. First, most existing methods based on fixed degradation suffer from significant performance drop when facing the unknown degradation, because of the huge gap between the fixed degradation and the unknown degradation. Second, the unknown-degradation estimation may lead to restoration task failure due to uncertain estimation errors. To handle the unknown degradation in the real application, we introduce a degradation representation network for single image blind restoration (DRN). Different from the methods of estimating pixel space, we use an encoder network to learn abstract representations for estimating different degradation kernels in the representation space. Furthermore, a degradation perception module with flexible adaptability to different degradation kernels is used to restore more structural details. In our experiments, we compare our DRN with several state-of-the-art methods for two image restoration tasks, including image super-resolution (SR) and image denoising. Quantitative results show that our degradation representation network is accurate and efficient for single image restoration.     

基于总变分正则化算法的散射计图像重构

目前业务化运行的星载散射计分辨率一般为25 km,在分辨率需求较高的应用中(如极地海冰监测、热带雨林监测、近岸风场研究等)受到了限制。散射计图像重构技术可以在不改变系统硬件的前提下,通过数据处理方法的改进,提高分辨率。现有的散射计图像重构方法(SIR)是基于图像处理领域中较早期的乘性代数重建技术(MART)。该文针对星载扇形旋转扫描散射计,将一种新的图像重构方法总变分正则化 (total variation regularization) 算法应用于散射计图像重构,并通过仿真实验说明,新算法可以在增强分辨率的同时减少噪声,提高重构图像的质量。     

Radon变换和全变分相融合的图像复原算法

图像复原的核心是点扩散函数的估计和直接去卷积算法,针对拍照过程中,相机和被拍摄物体由于相对运动而导致的图像退化问题,提出一种基于Radon变换和全变分相融合的图像复原算法。首先利用radon变换对图像退化模型参数进行估计,然后采用全变分算法复原退化图像,最后在Matlab 2012平台进行仿真实验对算法的性能检验。仿真结果表明,相对于其它图像复原算法,本文算法可以准确估计退化模型参数,获得了更加理想的图像复原效果,具有一定的实际利用价值。     

Image decomposition using optimally sparse representations and a variational approach

In this paper, a new method which combines the basis pursuit denoising algorithm (BPDN) and the total variation (TV) regularization scheme is presented for separating images into texture and cartoon parts. It is a modification of the model [1]. In this process, two appropriate dictionaries are used, one for the representation of texture parts-the dual tree complex wavelet transform (DT CWT) and the other for the cartoon parts-the second generation of curvelet transform. To direct the separation process and reduce the pseudo-Gibbs phenomenon, the curvelet transform is followed by a projected regularization method for cartoon parts. Experimental results show that new method cannot only decompose better for a given image but also reduce the runtime, in comparison to the MCA approach.     

Matrix-variate variational auto-encoder with applications to image process

Variational Auto-Encoder (VAE) is an important probabilistic technology to model 1D vectorial data. However, when applying VAE model to 2D image, vectorization is necessary. Vectorization process may lead to dimension curse and lose valuable spatial information. To avoid these problems, we propose a novel VAE model based on matrix variables named as Matrix-variate Variational Auto-Encoder (MVVAE). In this model, input, hidden and latent variables are all in matrix form, therefore inherent spatial structure of 2D images can be maintained and utilized better. Especially, the latent variable is assumed to follow matrix Gaussian distribution which is more suitable for describing 2D images. To solve the weights and the posterior of latent variable, the variational inference process is given. The experiments are designed for three real-world application: reconstruction, denoising and completion. The experimental results demonstrate that MVVAE shows better performance than VAE and other probabilistic methods for modeling and processing 2D data.     

MRI reconstruction using a joint constraint in patch-based total variational framework

Compressed sensing (CS) as an efficient means has been widely applied in magnetic resonance imaging (MRI). As a regularization term to enforce the sparsity in the finite difference domain, the conventional total variation (TV) has been introduced in this field, where the staircase effect is presented. To overcome this issue, a new framework in the difference domain called joint constraint patch-based total variation (JCTV) is proposed. First, the image patch is utilized as the unit for TV norm to improve the adaptativity. Second, JCTV introduces a new nonlocal constraint term that exploits the estimated coefficients of the fully sampled image via linear minimum mean square error (LMMSE) criterion to improve the reconstruction performance. Finally, an alternative minimization algorithm is developed to seek the solution. Extensive experiments on a set of in vivo MR images demonstrate that the proposed algorithm outperforms the state-of-the-art approaches in terms of peak signal-to-noise ratio and visual quality.