Fringe pattern denoising using averaging based on nonlocal self-similarity

Digital image processing for fringe patterns is an important procedure in optical interferometry. Filtering off noise from fringe patterns is one of the key tasks for extraction of the phase field. Spin filters proposed by Yu et al. [Appl. Opt. 33(1994), 41(2002), et al.] have been proven to be effective denoising methods. In this paper, we develop a nonlocal self-similarity filter, which averages similar pixels searched for in whole image instead of in a local fringe direction as the spin filters do. Although simple and free of the fringe orientation estimation, involving more pixels with higher similarity levels, our algorithm has stronger robustness against noise and thus denoises fringe patterns more effectively. Simulation and experimental results show that our algorithm outperforms related filters both in preserving smooth fringes and in reducing blurring effects and quantitative errors.     

Eliminating the zero spectrum in Fourier transform profilometry using a two-dimensional continuous wavelet transform

In this paper, a filtering technique based upon two-dimensional continuous wavelet transform (2D-CWT) is used to eliminate the low frequency components of fringe patterns. The filtered fringe patterns are subsequently demodulated using a standard Fourier transform profilometry (FTP) algorithm. This image pre-filtering stage improves the noise performance of the FTP algorithm and enables the FTP method to demodulate fringe patterns with larger bandwidths. Also, the 2D-CWT technique reduces speckle noise significantly. Moreover, only a single fringe pattern is required in this technique. The 2D-CWT algorithm is capable of separating low frequency terms from the high frequency terms that contain phase-modulated fringe information, even when both interfere, greatly, in the frequency domain. The proposed algorithm is tested, both via computer simulation and using real fringe patterns. This revealed the robustness of this algorithm and also demonstrably enables the demodulation of a wider range of fringe patterns using the FTP technique.     

基于改进旋滤波的电子散斑干涉图滤波方法

电子散斑干涉术条纹图在成像时不可避免地受散斑噪声调制,去除噪声是散斑干涉条纹处理的一项重要任务。利用散斑条纹图的方向性,提出一种基于模糊方向的旋滤波:在当前点的领域内定义4个模糊方向窗口,将传统旋滤波的一维、精确方向窗口的确定,转变为模糊方向窗口的确定;在确定的窗口内进行低通滤波时,采用自适应加权均值滤波代替传统的中值滤波。利用该方法分别处理模拟散斑条纹图和实验所得的真实条纹图,并与传统旋滤波、双边滤波和小波丢弃子带方法比较。实验结果表明,该改进算法在滤除散斑条纹图噪声的同时,有效保护了条纹的细节信息。     

散斑条纹图的自适应窗口滤波方法

针对散斑噪声很难用常用的滤波方法进行滤除,在已有等值线滤波方法的基础上,提出一种更优的自适应窗口滤波方法。由于等值线窗口是在条纹方向图的基础上得到的,深入研究条纹方向的求取方法,提出了一种更为可靠的条纹方向求取算法,在得到高精度条纹方向的同时,还能估计出条纹的密度。使用等值线窗口技术,保持了窗口形状的自适应能力,然后利用对条纹密度的估计,根据条纹的宽度来确定滤波窗口的大小,实现了滤波窗口大小的自适应。最后,根据条纹方向和滤波后的图像可以直接得到条纹密度变化较大的单幅散斑图的相位结果。     

Patch-wise denoising of phase fringe patterns based on matrix enhancement

We propose a new approach for the denoising of a phase fringe pattern recorded in an optical interferometric setup. The phase fringe pattern which is generally corrupted by high amount of speckle noise is first converted into an exponential phase field. This phase field is divided into a number of overlapping patches. Owing to the small size of each patch, the presence of a simple structure of the interference phase is assumed in it. Accordingly, the singular value decomposition (SVD) of the patch allows us to separate the signal and noise components effectively. The patch is reconstructed only with the signal component. In order to further improve the robustness of the proposed method, an enhanced data matrix is generated using the patch and the SVD of this enhanced matrix is computed. The matrix enhancement results in an increased dimension of the noise subspace which thus accommodates more amount of noise component. Reassignment of the filtered pixels of the preceding patch in the current patch improves the noise filtering accuracy. The fringe denoising capability in function of the noise level and the patch size is studied. Simulation and experimental results are provided to demonstrate the practical applicability of the proposed method.     

Anisotropic partial differential equation noise-reduction algorithm based on fringe feature for ESPI

Noise reduction is one of the most exciting problems in electronic speckle pattern interferometry. We present a new anisotropic partial differential equation noise-reduction algorithm based on fringe orientation for interferometric fringe patterns. The proposed equation performs diffusion along the two directions of fringe gradient and isophote line, which are extracted accurately according to fringe feature. By restriction of diffusion in the gradient direction of fringe patterns, this method can provide optimal results in denoising but does not destroy fringe edges. The experimental results show that this technique is more capable of significantly improving the quality of the fringe patterns than the classical anisotropic diffusion equation proposed by Perona and Malik. Based on our filtered fringe patterns, the phase map obtained by phase-shifting technique can be extracted more accurately. It is an effective pre-processing method for electronic speckle pattern interferometry.     

A weighted least squares algorithm for time-of-flight depth image denoising

In this paper, we present a new denoising method for the depth image of a time-of-flight (ToF) camera, based on weighted least squares (WLS) framework. The common method for ToF depth image denoising is to use bilateral filter. However, the ability of bilateral filter in edge preservation would be reduced while we attempt to smooth out larger spatial scale noise. In order to avoid this problem and preserve the edge information as much as possible, we introduce a new way to construct edge-preserving ToF depth image denoising based on WLS. We are to our knowledge the first to present a WLS-based method for ToF depth image denoising. Experimental results demonstrate that compared with bilateral filter, our proposed algorithm not only achieves better performance in edge preservation, but also improves the PSNR values of the denoised images by 0.5–2.6 dB.     

基于CCD采集的Mach-Zehnder干涉条纹图的处理算法

Mach-Zehnder干涉仪适用于研究气体密度迅速变化的状态。由于气体折射率的变化与其密度的变化成正比,而折射率的变化将使通过气体的光线有不同的光程,因此可通过干涉臂变化对干涉条纹图像效果的影响得到气体密度。实测中,采用图像采集卡和CCD来接收Mach-Zehnder干涉仪产生的条纹图像,再通过计算机对条纹图像的条纹间距进行处理,从而得到气体密度的变化状态。从光干涉理论出发,对Mach-Zehnder干涉条纹图像特征进行了分析,建立了Mach-Zehnder干涉条纹的数学模型,并根据此模型设计了处理Mach-Zehnder干涉条纹图像的算法。算法包括图像的预处理(即图像的噪声提取)、图像的二值化及图像的细化。     

Variational denoising method for electronic speckle pattern interferometry

Traditional speckle fringe patterns by electronic speckle pattern interferometry (ESPI) are inherently noisy and of limited visibility, so denoising is the key problem in ESPI. We present the variational denoising method for ESPI. This method transforms the image denosing to minimizing an appropriate penalized energy function and solving a partial differential equation. We test the proposed method on computer-simulated and experimental speckle correlation fringes, respectively. The results show that this technique is capable of significantly improving the quality of fringe patterns. It works well as a pre-processing for the fringe patterns by ESPI.     

散斑干涉相位条纹图的频域滤波处理

为了解决在数字散斑干涉技术测量时,散斑干涉相位条纹图像中大量噪声对相位解包裹结果和精度产生严重影响的问题,介绍了一种条纹正余弦分解和频域低通滤波结合的方法,实现了散斑干涉相位条纹图的高精度滤波。该方法的基本思路是在对相位图像进行滤波处理前,先将相位图通过正余弦函数进行映射转换成两幅图,分别经过频域滤波,然后再合成为相位图。这种分解频域滤波方法可以在滤波的同时,有效保留相位跳变信息。实验结果表明：与传统的图像降噪方法相比,该方法能够在保留图像“尖峰”信息的基础上,较好地滤除图像中的散斑噪声,方法简单有效,有效解决了传统滤波方法应用在相位条纹图中,相图灰度信息丢失10%~40%的问题。     

基于全卷积金字塔残差网络的能谱CT图像降噪研究

传统CT采用积分式探测器采集投影数据,反映的是物体的平均衰减特性,会在一定程度上造成信息损失,无法对物体进行较好的定性定量测量。基于光子计数探测器的能谱CT通过设定多个能量响应阈值能够探测不同能量范围内的X射线光子,采集更多被测物体的物质组成信息,有助于识别不同物理特性的材料,基于此,能谱CT被广泛的应用于小病灶、低对比度结构以及微细结构的成像。然而将整个能谱划分为多个能量段进行数据采集时,范围较窄能量范围内的有效光子数比例相对降低,导致图像中包含较多的噪声,图像质量较差,影响能谱CT的临床应用。为了有效的抑制能谱CT不同能量段内重建图像中的噪声,提出了一种基于深度学习的能谱CT图像降噪方法。我们将全卷积网络和金字塔残差网络结合为全卷积金字塔残差网络(FCPRN),实验中,利用能谱CT在不同的能量范围扫描小鼠样本,使用FDK算法和基于压缩感知的Split-Bregman算法进行重建并分别作为训练数据和标签数据训练全卷积金字塔残差网络。为了验证网络的降噪性能,选取了常见的降噪网络模型denoising convolutional neural networks(DNCN)以及residual encoder decoder convolutional neural network (REDCNN)进行对比,训练三种网络的使用的数据和实验配置都是完全相同的,实验结果表明训练模型可以有效抑制不同能量范围内重建图像的噪声,且使用的全卷积金字塔残差网络的降噪性能优于其他网络模型。模型训练好后,可以对FDK算法重建出的图像进行降噪,由此提高能谱CT图像降噪效率,保证能谱CT重建图像的质量。     

Generalized total variation-based MRI Rician denoising model with spatially adaptive regularization parameters

Magnetic resonance imaging (MRI) is an outstanding medical imaging modality but the quality often suffers from noise pollution during image acquisition and transmission. The purpose of this study is to enhance image quality using feature-preserving denoising method. In current literature, most existing MRI denoising methods did not simultaneously take the global image prior and local image features into account. The denoising method proposed in this paper is implemented based on an assumption of spatially varying Rician noise map. A two-step wavelet-domain estimation method is developed to extract the noise map. Following a Bayesian modeling approach, a generalized total variation-based MRI denoising model is proposed based on global hyper-Laplacian prior and Rician noise assumption. The proposed model has the properties of backward diffusion in local normal directions and forward diffusion in local tangent directions. To further improve the denoising performance, a local variance estimator-based method is introduced to calculate the spatially adaptive regularization parameters related to local image features and spatially varying noise map. The main benefit of the proposed method is that it takes full advantage of the global MR image prior and local image features. Numerous experiments have been conducted on both synthetic and real MR data sets to compare our proposed model with some state-of-the-art denoising methods. The experimental results have demonstrated the superior performance of our proposed model in terms of quantitative and qualitative image quality evaluations.     

一种基于双分支改良编解码器的图像去噪算法

针对传统图像去噪算法多噪声去除难,深层卷积神经网络去噪模型网络复杂、训练时间长等问题,提出一种基于自编码器结构的双分支改良编解码网络,实现高效图像去噪。双分支结构之一采用降-升采样实现点噪声消除,另一分支专注于宏观的图像修复和伪像去除,后端利用残差结构进行整合,实现数字图像混合噪声去噪。实验结果显示:对于含有标准差为15,均值为0的高斯噪声、噪声密度为5%的椒盐噪声和散粒噪声的混合噪声图像测试集,实验去噪效果相较于输入混合噪声图像峰值信噪比,平均提升了5.3%。与12层全卷积神经网络相比,去噪效果相当,训练速度提升了25.4%,体现了其“轻量级”的优点。实验表明:该方法相较于深层卷积神经网络,训练速度快,网络简单;相较于传统图像去噪算法,噪声去除效果也较为明显。该算法可应用于轻量级视觉平台后端去噪。     

基于条纹等值线相关的散斑条纹图处理新方法

在数字散斑干涉术中,通常由物体变形前后得到的两幅原始散斑场用相减方法来得到条纹图,并由相移后的多幅(三幅以上)条纹图来求取相位场。介绍了一种基于条纹等值线相关的条纹图生成新方法,并在此基础上提出了基于单幅干涉条纹图的相位场提取新方法和基于条纹等值线相关的单步相移新方法。实验结果表明,该方法具有良好的结果和发展前景。     

Piston sensing via a dispersed fringe sensor with a merit-function-based active scanning algorithm at low light levels

Dispersed fringe sensors are a promising approach for sensing the large-scale physical step between adjacent segments with acceptable accuracy. However, the nature of dispersion in a dispersed fringe sensor leads to the ideal dispersed fringe pattern becoming vulnerable to noise, particularly at low light levels. A reliable merit-functionbased algorithm with an active actuation is introduced here. The feasibility of our algorithm is numerically demonstrated, and Monte Carlo experiments for different signal-to-noise ratios are conducted to assess its robustness. The results show that the method is valid even when the signal-to-noise ratio is as low as 1.     

一种基于信息保留网络的图像去噪算法

由于成像设备等各种因素影响, 图像在成像或传感过程中会受到噪声干扰。图像去噪旨在减少或消除噪声对图像的影响, 这一过程往往会导致高频信息的丢失。为了在去除图像噪声的同时保护图像的边缘信息与纹理细节, 文章提出了一种计算复杂度相对较低的含有信息保留模块的卷积神经网络, 直接对含噪声图像进行降噪。信息保留模块通过残差学习提取局部长路径和局部短路径的混合特征信息。该文采用峰值信噪比(PSNR/dB)和结构相似性(SSIM)两项评价指标对实验结果进行量化, 这两项指标值越大, 说明去噪效果越好。实验结果表明, 在峰值信噪比和结构相似性2项评价指标的均值可达到30.36 dB和0.828 0, 相比其他对比算法, 2项评价指标分别平均提升了2.15 dB和0.072 9。该算法对不同种类、不同水平的噪声都具有良好的去噪效果, 且速度优于所对比的一般算法, 对基于卷积神经网络的去噪工作的进一步发展有一定的作用。     

混沌信号自适应协同滤波去噪

混沌信号协同滤波去噪算法充分利用了混沌信号的自相似结构特征,具有良好的信噪比提升性能.针对该算法的滤波参数优化问题,考虑到最优滤波参数的选取受到信号特征、采样频率和噪声水平的影响,为提高该算法的自适应性使其更符合实际应用需求,基于排列熵提出一种滤波参数自动优化准则.依据不同噪声水平的混沌信号排列熵的不同,首先选取不同滤波参数对含噪混沌信号进行去噪,然后计算各滤波参数对应重构信号的排列熵,最后通过比较各重构信号的排列熵,选取排列熵最小的重构信号对应的滤波参数为最优滤波参数,实现滤波参数的优化.分析了不同信号特征、采样频率和噪声水平情况下滤波参数的选取规律.仿真结果表明,该参数优化准则能在不同条件下对滤波参数进行有效的自动最优化,提高了混沌信号协同滤波去噪算法的自适应性.     

A novel color fringe projection based Fourier transform 3D shape measurement method

A novel Fourier transform 3D shape measurement method based on color fringe projection is proposed in order to solve the spectrum overlapping and phase unwrapping problems existed in Fourier transform profilometry (FTP). The R and G components of the color fringe are set to two sinusoidal patterns with different frequencies and the B component is set to the average value of R or G component. Then this pattern is projected to the object and the deformed fringe image is captured. Three gray patterns are separated from the color fringe, the background and high frequency noise can be eliminated using our method and the accurate unwrapped phase can be got. Only one shot color pattern is projected to get the 3D information of the object. Experiment results show that the 3D information of an object can be obtained rapidly and accurately.     

A novel denoising method based on Radon transform and filtered back-projection reconstruction algorithm

A novel denoising method based on Radon transform and filtered back-projection (FBP) image reconstruction algorithm was proposed. This method can be considered as a special mean filter on projection line, which is different from most of the traditional filters operated on adjacent templates that could bring serious blurs to images. The details of images processed by the proposed method can be preserved relatively complete and the denoising effect is satisfactory. To verify the denoising effect of the proposed method, the simulation was designed and carried out, and the image evaluation parameters were applied to analyze the denoising effect and the detail-preserving ability quantitatively. For further understanding of the proposed method, the basic denoising principle of this method was analyzed. Noise points and information points can be distinguished: the attenuation velocity of gray scale of noise points is faster than that of information points, which was verified by the experiment. The results of different parameters in the proposed method were compared and analyzed. Several kinds of traditional filters were compared with the proposed method, and the result shows that the proposed method is better than the traditional filters in the aspects of both denoising effect and detail-preserving ability. Apart from this, the proposed method is not particular about the kind of noise; therefore, it is a powerful method to deal with atypical noise, uncertain noise, and mixed noises.     

基于结构字典学习的图像复原方法

本文提出一种新的结构字典学习方法,并利用它进行图像复原。首先给出结构字典学习的基本内容和方法,然后将傅里叶正则化方法和结构字典学习方法有效整合到图像复原算法中。结构字典学习方法是先将原图像进行结构分解,再分别学习出每个结构图像中的字典,最后利用这些字典对原图像进行稀疏的表示。结合傅里叶正则化,提出了一种有效的迭代图像复原算法:第一步在傅里叶域利用正则化反卷积方法得到图像的初步估计;第二步采用结构字典学习的方法对遗留的噪声进行去噪处理。实验结果表明,提出的方法在改进信噪比和视觉质量上都要优于6种先进的图像复原方法,改进的信噪比平均提升0.5 d B以上。