Wavelet inverse scale space for image restoration

This paper proposes a model for image restoration by combining the wavelet shrinkage and inverse scale space （ISS） method. The ISS is applied to the wavelet representation to modify the retained wavelet coefficients, and the coefficients smaller than the threshold are set to zero. The curvature term of the ISS can remove the edge artifacts and preserve sharp edges. For the multiscale interpretation of the ISS and the multiscale property of the wavelet representation, small details are preserved. This paper illustrates that the wavelet ISS model can be deduced from the wavelet based on a total variation minimization problem. A stopping criterion is obtained from this minimization in the sense of the Bregman distance in the wavelet domain. Numerical examples show the improvement for the image denoising with the proposed method in the sense of the signal to noise ratio and with fewer details remained in the residue.

A METHOD TO APPROACH OPTIMAL RESTORATION IN IMAGE RESTORATION PROBLEMS WITHOUT NOISE ENERGY INFORMATION

This paper proposes a new image restoration technique, in which the resulting regularized image approximates the optimal solution steadily. The affect of the regular-ization operator and parameter on the lower band and upper band energy of the residue of the regularized image is theoretically analyzed by employing wavelet transform. This paper shows that regularization operator should generally be lowstop and highpass. So this paper chooses a lowstop and highpass operator as regularization operator, and construct an optimization model which minimizes the mean squares residue of regularized solution to determine regularization parameter. Although the model is random, on the condition of this paper, it can be solved and yields regularization parameter and regularized solution. Otherwise, the technique has a mechanism to predict noise energy. So, without noise information, it can also work and yield good restoration results.

THE SINE TRANSFORM OPERATOR IN THE BANACH SPACE OF SYMMETRIC MATRICES AND ITS APPLICATION IN IMAGE RESTORATION

In this paper, we study an operator s which maps every n-by-n symmetric matrix A, to a matrix s(A_n) that minimizes || B_n-A_n || F over the set of all matrices B_n, that can be diagonalized by the sine transform. The matrix s(A_n), called the optimal sine transform preconditioner, is defined for any n-by-n symmetric matrices A_n. The cost of constructing s(A_n) is the same as that of optimal circulant preconditioner c(A_n) which is defined in [8], The s(A_n) has been proved in [6] to be a good preconditioner in solving symmetric Toeplitz systems with the preconditioned conjugate gradient (PCG) method. In this paper, we discuss the algebraic and geometric properties of the operator s, and compute its operator norms in Banach spaces of symmetric matrices. Some numerical tests and an application in image restoration are also given.

Wavelet-Based Diffusion Approach for DTI Image Restoration

The Rician noise introduced into the diffusion tensor images （DTIs） can bring serious impacts on tensor calculation and fiber tracking. To decrease the effects of the Rician noise, we propose to consider the wavelet-based diffusion method to denoise multiehannel typed diffusion weighted （DW） images. The presented smoothing strategy, which utilizes anisotropic nonlinear diffusion in wavelet domain, successfully removes noise while preserving both texture and edges. To evaluate quantitatively the efficiency of the presented method in accounting for the Rician noise introduced into the DW images, the peak-to-peak signal-to-noise ratio （PSNR） and signal-to-mean squared error ratio （SMSE） metrics are adopted. Based on the synthetic and real data, we calculated the ap- parent diffusion coefficient （ADC） and tracked the fibers. We made comparisons between the presented model, the wave shrinkage and regularized nonlinear diffusion smoothing method. All the experiment results prove quantitatively and visually the better performance of the presented filter.

Dark channel prior based blurred image restoration method using total variation and morphology

The blurred image restoration method can dramatically highlight the image details and enhance the global contrast, which is of benefit to improvement of the visual effect during practical ap- plications. This paper is based on the dark channel prior principle and aims at the prior information absent blurred image degradation situation. A lot of improvements have been made to estimate the transmission map of blurred images. Since the dark channel prior principle can effectively restore the blurred image at the cost of a large amount of computation, the total variation （TV） and image morphology transform （specifically top-hat transform and bottom- hat transform） have been introduced into the improved method. Compared with original transmission map estimation methods, the proposed method features both simplicity and accuracy. The es- timated transmission map together with the element can restore the image. Simulation results show that this method could inhibit the ill-posed problem during image restoration, meanwhile it can greatly improve the image quality and definition.

An overview of image restoration based on variational regularization

Image restoration is a complicated process in which the original information can be recovered from the degraded image model caused by lots of factors.Mathematically,image restoration problems are ill-posed inverse prob-lems.In this paper image restoration models and algorithms based on variational regularization are surveyed.First,we review and analyze the typical models for denoising,deblurring and inpainting.Second,we construct a unified restoration model based on variational regularization and summarize the typical numerical methods for the model.At last,we point out eight diffcult problems which remain open in this field.

Radon transform-based motion blurred silkworm pupa image restoration

As for machine vision-based intelligent system in the application of discriminating and sorting the sex of silkworm pupae,the tail gonad was the unique physiological feature.However,motion blur,resulting from the live silkworm pupa’s writhing motion at the moment of capturing image,could lose textures and structures(such as edge and tail gonad etc.)dramatically,which casted great challenges for sex identification.To increase the image quality and relieve the difficulty of discrimination caused by motion blur,an effective approach that including three stages was proposed in this work.In the image prediction stage,first sharp edges were acquired by using filtering techniques.Then the initial blur kernel was computed with Gaussian prior.The coarse version latent image was deconvoluted in the Fourier domain.In the kernel refinement stage,the Radon transform was applied to estimate the accurate kernel.In the final restoration step,a TV-L1 deconvolution model was carried out to obtain a better result.The experimental results showed that benefiting from the prediction step and kernel refinement step,the kernel was more accurate and the recovered image contained much more textures.It revealed that the proposed method was useful in removing the motion blur.Furthermore,the method could also be applied to other fields.

Diffusion filtering in image processing based on wavelet transform

The nonlinear diffusion filtering in image processing bases on the heat diffusion equations. Its key is the control of diffusion amount. In the previous models, the diffusivity depends on the gradients of images. So it is easily affected by noises. This paper first gives a new multiscale computational technique for diffusivity. Then we proposed a class of nonlinear wavelet diffusion （NWD） models that are used to restore images. The NWD model has strong ability to resist noise. But it, like the previous models, requires higher computational effort. Thus, by simplifying the NWD, we establish linear wavelet diffusion （LWD） models that consist of advection and diffusion. Since there exists the advection, the LWD filter is anisotropic, and hence can well preserve edges although the diffusion at edges is isotropic. The advantage is that the LWD model is easy to be analyzed and has lesser computational load. Finally, a variety of numerical experiments compared with the previous model are shown.

UAV remote sensing atmospheric degradation image restoration based on multiple scattering APSF estimation

Unmanned aerial vehicle(UAV) remote imaging is affected by the bad weather, and the obtained images have the disadvantages of low contrast, complex texture and blurring. In this paper, we propose a blind deconvolution model based on multiple scattering atmosphere point spread function(APSF) estimation to recovery the remote sensing image. According to Narasimhan analytical theory, a new multiple scattering restoration model is established based on the improved dichromatic model. Then using the L0 norm sparse priors of gradient and dark channel to estimate APSF blur kernel, the fast Fourier transform is used to recover the original clear image by Wiener filtering. By comparing with other state-of-the-art methods, the proposed method can correctly estimate blur kernel, effectively remove the atmospheric degradation phenomena, preserve image detail information and increase the quality evaluation indexes.

A NEW APPROACH FOR UNSUPERVISED RESTORING IMAGES BASED ON WAVELET-DOMAIN PROJECTION PURSUIT LEARNING NETWORK

The Wavelet-Domain Projection Pursuit Learning Network (WDPPLN) is proposedfor restoring degraded image. The new network combines the advantages of both projectionpursuit and wavelet shrinkage. Restoring image is very difficult when little is known about apriori knowledge for multisource degraded factors. WDPPLN successfully resolves this problemby separately processing wavelet coefficients and scale coefficients. Parameters in WDPPLN,which are used to simulate degraded factors, are estimated via WDPPLN training, using scalecoefficients. Also, WDPPLN uses soft-threshold of wavelet shrinkage technique to suppress noisein three high frequency subbands. The new method is compared with the traditional methodsand the Projection Pursuit Learning Network (PPLN) method. Experimental results demonstratethat it is an effective method for unsupervised restoring degraded image.

基于SVC和wavelet-transform的图像脉冲噪声自适应新滤波器

利用小波变换可以检测信号奇异点的原理,提出了一种基于WT的脉冲噪声检测方法,并把这一方法与支持向量分类器SVC脉冲噪声检测方法相结合,提出了一种改进的SVC图像脉冲噪声滤波器。实验表明,这一改进的SVC脉冲噪声滤波器的滤波效果比原先的SVC滤波器有明显的改善。

基于Transformer的陶瓷轴承表面缺陷检测方法

针对传统机器视觉检测方法中,由于陶瓷轴承滚动体表面曲率大、对比度低,表面成像模糊导致后续缺陷检测精度低的问题,提出一种基于Transformer的超分辨率残差网络。首先,网络使用残差学习策略,通过预测模糊图像与清晰图像之间的差值,实现超分辨率任务;其次,在网络上前端插入通道注意力模块和空间注意力模块并改进L2多头自注意力模块,以增强图像纹理、改善梯度爆炸问题;最后,针对超分辨率重建任务,提出一种两阶段训练策略优化训练过程。自建陶瓷轴承表面缺陷数据集上的大量实验结果表明,所提出网络模型在客观指标与主观评价上均优于MSESRGAN、VSDR等超分辨率算法,重建图像SSIM为0.939,PSNR为36.51 dB。

New Edge-Directed Interpolation Based-Lifting DWT and MSPIHT Algorithm for Image Compression

The amount of image data generated in multimedia applications is ever increasing. The image compression plays vital role in multimedia applications. The ultimate aim of image compression is to reduce storage space without degrading image quality. Compression is required whenever the data handled is huge they may be required to sent or transmitted and also stored. The New Edge Directed Interpolation (NEDI)-based lifting Discrete Wavelet Transfrom (DWT) scheme with modified Set Partitioning In Hierarchical Trees (MSPIHT) algorithm is proposed in this paper. The NEDI algorithm gives good visual quality image particularly at edges. The main objective of this paper is to be preserving the edges while performing image compression which is a challenging task. The NEDI with lifting DWT has achieved 99.18% energy level in the low frequency ranges which has 1.07% higher than 5/3 Wavelet decomposition and 0.94% higher than traditional DWT. To implement this NEDI with Lifting DWT along with MSPIHT algorithm which gives higher Peak Signal to Noise Ratio (PSNR) value and minimum Mean Square Error (MSE) and hence better image quality. The experimental results proved that the proposed method gives better PSNR value (39.40 dB for rate 0.9 bpp without arithmetic coding) and minimum MSE value is 7.4.

Satellite Image Adaptive Restoration Using Periodic Plus Smooth Image Decomposition and Complex Wavelet Packet Transforms

A satellite image adaptive restoration method was developed that avoids ringing artifacts at the image boundary and retains oriented features. The method combines periodic plus smooth image decom- position with complex wavelet packet transforms. The framework first decomposes a degraded satellite im- age into the sum of a ＂periodic component＂ and a ＂smooth component＂. The Bayesian method is then used to estimate the modulation transfer function degradation parameters and the noise. The periodic component is deconvoluted using complex wavelet packet transforms with the deconvolution result of the periodic component then combined with the smooth component to get the final recovered result. Tests show that this strategy effectively avoids ringing artifacts while preserving local image details （especially directional tex- tures） without amplifying the noise. Quantitative comparisons illustrate that the results are comparable with previous methods. Another benefit is that this approach can process large satellite images with parallel processing, which is important for practical use.

基于Transformer的文物图像修复方法

文物极易因为保存不当而导致部分结构或纹理缺失,而现有的图像修复技术由于受到先验信息和卷积操作的局限而无法直接应用于文物图像修复,为更合理地恢复文物图像原貌,提出了一种新的文物图像修复方法,将文物图像修复工作分为2个步骤:第1步使用Transformer进行粗略的图像重建并恢复连贯的结构;第2步使用卷积神经网络将粗略的重建图像进行上采样并恢复缺失区域的精细纹理.考虑到目前国内外没有高质量的大型文物数据库,因此也提出了一个新的高质量大型文物图像数据库.最终实验结果表明,在符合现实场景的破损修复实验和大面积破损修复实验中,修复效果在主观和客观评估中均优于当前图像修复算法.同时,支持多元化输出,为修复人员提供多样化参考,极大地提升了文物修复效率.

R-L Method and BLS-GSM Denoising for Penumbra Image Reconstruction

When neutron yield is very low, reconstruction of coding penumbra image is rather difficult. In this paper, low-yield （109） 14 MeV neutron penumbra imaging was simulated by Monte Carlo method. The Richardson Lucy （R-L） iteration method was proposed to incorporated with Bayesian least square-Gaussian scale mixture model （BLS-GSM） wavelet denoising for the simulated image. Optimal number of R-L iterations was gotten by a large number of tests. The results show that compared with Wiener method and median filter denoising, this method is better in restraining background noise, the correlation coefficient Rsr between the reconstructed and the real images is larger, and the reconstruction result is better.

结合梯度指导和局部增强Transformer的图像去模糊网络

模糊图像不仅影响人类感知还会影响后续计算机视觉任务的性能,例如自动驾驶系统和户外监控系统中的视觉算法.针对以往基于深度学习的去模糊方法感受野较小,不能动态适应输入内容和重建图像细节信息困难等问题,提出了一种基于Transformer的图像去模糊网络.网络包含两个分支:图像内容分支和梯度分支,每条分支均以具有窗口机制的Transformer作为主干,通过梯度分支的信息指导图像去模糊重建,能够更好地恢复图像的边缘和纹理.同时,为了充分利用图像的内容信息和梯度信息,本文还设计了一个交互式融合模块来有效融合特征信息.此外,本文通过在Transformer块的自注意力机制和前馈网络中引入卷积来解决Transformer对局部信息建模不足的问题.在合成数据集和真实数据集上的大量实验结果表明,提出的算法能有效去除复杂模糊并且恢复清晰的细节,在定量指标和视觉效果上均优于目前的主流去模糊算法.

基于Transformer的多阶段运动模糊图像修复网络

运动模糊是导致图像退化的常见原因,其限制了图像的可读性和后续处理效果。针对卷积网络感受野有限以及常规多阶段网络中信息丢失的问题,提出一种基于Transformer的多阶段去模糊网络。网络采用多阶段编码器-解码器结构,在单个阶段内和多个阶段间采用跳跃连接来增强信息的传递。首先,高效Transformer模块采用通道注意力和深度卷积来处理图像的全局和局部信息;其次,多分支结构的前馈传播网络通过引入多个并行的分支,实现了不同尺度和不同层次的特征提取和融合;最后,通过多阶段的残差处理实现更优的图像恢复结果。实验结果显示,在GoPro数据集上该网络的峰值信噪比(PSNR)达到32.23 dB,结构相似性指数(SSIM)达到0.955,在HIDE数据集上PSNR和SSIM分别达到30.15 dB和0.930,优于DeepDeblur、DeblurGAN-V2等模型。

面向图像复原和增强的轻量级交叉门控Transformer

现有的图像复原和图像增强方法难以同时兼顾在多个子任务上的鲁棒性和维持较小的参数量与计算代价。针对这一问题,提出轻量级交叉门控转换算法(CGT)。一方面,总结了传统全局自注意力机制捕获全局依赖关系的局限性,将全局自注意力机制改进为跨层次交叉门控自注意力机制。同时提出轻量化的前馈神经网络,从而以极小的计算代价学习到跨层次局部依赖关系,在局部邻域内重构清晰特征。另一方面,针对传统方法对编码器和解码器平等地进行加法或拼接的操作易导致信息干扰这一缺陷,提出长距离重置更新模块,分别对无用信息与清晰特征加以抑制和更新。在图像去噪、图像去雨和低亮度图像增强3个不同任务的9个公开数据集上,与最新的25个方法进行的对比实验结果表明,所提出的轻量级交叉门控转换模型以较少的参数量和计算代价,在图像复原和图像增强领域中均取得较高的峰值信噪比和结构相似度,重构出接近真实世界场景的清晰图像,达到了先进的图像复原性能。

An Improved SPIHT Algorithm for Image Compression in Low Bit Rate

Aiming at shortage of the SPIHT algorithm, an improved image compression algorithm is proposed, in order to overcome the shortcomings of decoding image quality and coding time, LS9/7 lifting wavelet transform is adopted. According to the characteristics of the human visual system (HVS), the scanning mode and the method to determine the threshold of algorithm are changed to improve the quality of reconstruction image. On the question of repeating scan of SPIHT algorithm, using maximum list thought, greatly reduce the computation and save operating time. The experimental results have proved that the improved algorithm of image decoding time and the quality of reconstruction images are better than the original algorithm , especially in the case of low bit rate.