Denoising Research Articles

Convex Relaxation of Vectorial Problems with Coupled Regularization

We propose convex relaxations for nonconvex energies on vector-valued functions which are tractable yet as tight as possible. In contrast to existing relaxations, we can handle the combination of nonconvex data terms with coupled regularizers such as $l^2$-regularizers. The key idea is to consider a collection of hypersurfaces with a relaxation that takes into account the entire functional rather than separately treating the data term and the regularizers. We provide a theoretical analysis, detail the implementations for different functionals, present run time and memory requirements, and experimentally demonstrate that the coupled $l^2$-regularizers give systematic improvements regarding denoising, inpainting, and optical flow estimation.

A Survey of Fuzzy Based Image Denoising Techniques

A Survey of Fuzzy Based Image Denoising Techniques

Analysis Electroencephalogram Signals Using ANFIS and Periodogram Techniques

In this paper the applications the Adaptative Neuro Fuzzy Inference Systems (ANFIS), the Empirical Mode Decomposition (EMD) and Discrete Wavelet Distribution (DWT) are used. An electroencephalogram (EEG) is a diagnostic test which measures the electrical activity of the brain using highly sensitive recording equipment attached to the scalp by fine electrodes. An EEG recording is often affected with noises. These noises strongly affect the visual analysis of EEG. To overcome this problem the denoising techniques as ANFIS, EMD and DWT are applied. The efficiency of the ANFIS, EMD and DWT to remove the noises was evaluated by several standard metrics between filter EEG output and clean original signal. The results obtained show that the ANFIS outperformed other denoising techniques in terms of localization of the components of the abnormal EEG signal. Due to non-stationary nature of the EEG signal, the uses of time-frequency techniques are inevitable. The parametric time-frequency technique used is Periodogram (PE). The EEG signals used are normal and abnormal; the abnormal signals are obtained from the patient that has the sleep-disordered breathing (SDB) and the patient that has the sleep movement disorders (periodic leg movements or PLM). The PE technique shows its higher performance at the level of resolution and deleting any interference-terms over other non-parametric time-frequency techniques given in the scientific literature. This study demonstrates that the combination of ANFIS and the PE techniques are a good issue in the in biomedicine. For experimental study we have used the MIT/BIH arrhythmia database. Simulations were carried out in MATLAB environment.

An Efficient Approach for Denoising of CT-Images Using EMD and Dual Tree Complex Wavelet Packets

Computed tomography (CT) images are generally corrupted by several noises from the measurement process complicating the automatic feature extraction and analysis of clinical data. To achieve the best possible diagnosis it is important that medical images be sharp, clear, and free of noise and artifacts. While the technologies for acquiring digital medical images continue to improve, resulting in images of higher and higher resolution and quality, noise remains an issue for many medical images. Removing noise in these digital images remains one of the major challenges in the study of medical imaging. A variety of literatures have been developed to solve the problem of medical images denoising which is a significant stage in an automatic diagnosis system. In this paper, we propose a new image denoising technique using EMD and Dual Tree Complex Wavelet Packets. Here, histon process is used in order to overcome the smoothing filter type and it will not affect the lower dimensions. We have used two noises, like as Gaussian and salt & pepper for the proposed technique. The performance of the proposed image denoising technique is evaluated on the five CT images using the PSNR and SDME. For comparison analysis, our proposed denoising technique is compared with the existing work in various noise levels. From the results, we can conclude that the proposed denoising technique has shown the SDME of 48.33 but the existing technique show the PSNR of 39.84 for salt & pepper noise.

An Improved Image Denoising Approach Using Optimized Variance-Stabilizing Transformations

Image denoising plays a very important process in image processing. Many researchers would study about the various noise removal approaches used for fluorescence images. The fluorescence images mainly contain Poisson noise. The process of denoising is the image is gaussianized first then the resulting image is given as an input to the OWT SURELET to remove the Gaussian white noise. The OWT SURELET is one of the conventional denoising algorithms used to remove the noises. Then to attain an exact signal an inverse transform is applied to the denoised signal. Difficulties may arise to choose the inverse transformation for fluorescence images because the bias error may occur if the non linear forward transform is applied. To overcome these difficulties a study is made on the proposed Anscombe transformation and OWT-SURELET suitable for fluorescence images. The proposed OWT-SURELET is compared with BLS_GSM strategy and the results are discussed. Experimental result shows that the proposed system is more efficient than the existing system, the results are tested using ISNR changes with the denoising algorithms and the inverse transforms.

A New Approach of Wavelet De-Noising on Various Types of Acoustic Emission Partial Discharges Signals

Recovering the detected acoustic emission of partial discharge (PD) from noise interference is essential in determining its type. Various researches have shown that the wavelet has the potential in eliminating noise interference in acoustic emission of PD signal. This work describes a new de-noising method by employing the new thresholding approach using the present available structure of wavelet transform and its components. Experimental works is performed in order to obtain acoustic signal from three types of PD namely corona discharge, internal discharge and surface discharge. A critical assessment is carried out in order to ascertain the capability of the proposed method. The analysis using the mother wavelet and its frequency bands are then investigated in order to identify the best selection for the de-noising selection. The achieved results are measured based on their level of cleanliness. From the results, it can be concluded that the selected structure for the new method using the proposed method shows a significant improvement and excelled by 25.1% against the compared method in de-noising the acoustic emission of PD signals

A New Dual Tree Wavelet Based Image Denoising Using Fuzzy Shrink and Lifting Scheme

Images are often corrupted by noise owing to channel communication errors, defective image acquisition plans or devices, engine sparks, power interference and atmospheric electrical emissions. In proposed method a new wavelet shrinkage algorithm based on fuzzy logic and the lifting scheme is used. In particular, intra-scale dependency within wavelet coefficients is modeled using a fuzzy feature. This fuzzy feature differentiates between important coefficients, namely, image discontinuity coefficients and noisy coefficients. The same is used for enhancing wavelet coefficients' information in the shrinkage step, which results in the fuzzy membership function shrinking the wavelet coefficients based on the fuzzy feature. Examine that the proposed image denoising algorithm in the new shiftable and modified version of discrete wavelet transform, is the dual-tree discrete wavelet transform. Also that the lifting scheme is there it allows a faster implementation of the WT and a fully in-place calculation of the WT. In addition, no extra memory is needed and the original signal can be replaced with its WT. The method transforms an image into the wavelet domain using lifting based wavelet filters in the wavelet domain and finally transforms the result into the spatial domain and the fuzzy threshold is used to extract out the Speckle in highest subbands. Experimental Result shows that the resultant enhanced image obtained has a better performance in noise suppression and edge preservation.

噪音工业计算机断层图像中追踪三维边缘曲面

A 3Dedge surface denoised tracking algorithm is proposed to reconstruct high accuracy edge surfaces from the noisy industrial CT slices based on the 3Dfractional-order integral.The 2Dfractional-order integral method has effective denoising ability to preserves the texture detail of the image,and it has low computation complexity and easy implementation due to the filtering mask.In this paper,the 2Dfractional-order integral has been extended to three-dimensional images,its 3Dcontinuous theory and the discrete filtering masks are also proposed,we call it volumetric fractional-order integral.Since the Laplacian operator shows the sensitivity to the noise,the traditional 3Dedge surface tacking method cannot extract the high precision 3Dedge surface from noisy slice images effectively,the 3Dfractional-order integral is added to the tracking method to overcome the existed shortcoming.Our method is able to detect and extract the 3Dedge surface of sub-voxel accuracy from the 2Dnoisy industrial CT slice images.The experiments have reported very encouraging results according to signal noise ratio and visual effect by comparing it to the tacking method based on 3DGaussian denosing method.

A novel method to determine the redshifts of active galaxies based on wavelet transform

Automatically determining redshifts of galaxies is very important for astronomical research on large samples, such as large-scale structure of cosmological significance. Galaxies are generally divided into normal galaxies and active galaxies, and the spectra of active galaxies mostly have more obvious emission lines. In the present paper, the authors present a novel method to determine spectral redshifts of active galaxies rapidly based on wavelet transformation mainly, and it does not need to extract line information accurately. This method includes the following steps: Firstly, we denoised a spectrum to be processed; Secondly, the low-frequency spectrum was extracted based on wavelet transform, and then we could get the residual spectrum through the denoised spectrum subtracting the low-frequency spectrum; Thirdly, the authors calculated the standard deviation of the residual spectrum and determined a threshold value T, then retained the wavelength set whose corresponding flux was greater than T; Fourthly, according to the wavelength form of all the standard lines, we calculated all the candidate redshifts; Finally, utilizing the density estimation method based on Parzen window, we determined the redshift point with maximum density, and the average value of its neighborhood would be the final redshift of this spectrum. The experiments on simulated data and real data from SDSS-DR7 show that this method is robust and its correct rate is encouraging. And it can be expected to be applied in the project of LAMOST.

Contourlet transform as an effective method for agricultural product image denoising

Contourlet transform as an effective method for agricultural product image denoising

An Approach to Wavelet Based Image Denoising

An Approach to Wavelet Based Image Denoising

An Improved Anisotropic Diffusion Model Based on Wavelet Transform for Image Denosing

Preserving meaningful details such as blurred thin edges and low-contrast fine features is important in image de noising. A new method based on improved anisotropic diffusion model and wavelet transform is presented for image denoising. The proposed diffusion model incorporates both local gradient and gray-level variance to preserve edges and fine details while effectively removing noise in low-contrast surface images.

On the importance of mathematical methods for analysis of MALDI-imaging mass spectrometry data.

In the last decade, matrix-assisted laser desorption/ionization (MALDI) imaging mass spectrometry (IMS), also called as MALDI-imaging, has proven its potential in proteomics and was successfully applied to various types of biomedical problems, in particular to histopathological label-free analysis of tissue sections. In histopathology, MALDI-imaging is used as a general analytic tool revealing the functional proteomic structure of tissue sections, and as a discovery tool for detecting new biomarkers discriminating a region annotated by an experienced histologist, in particular, for cancer studies. A typical MALDI-imaging data set contains 10⁸ to 10⁹ intensity values occupying more than 1 GB. Analysis and interpretation of such huge amount of data is a mathematically, statistically and computationally challenging problem. In this paper we overview some computational methods for analysis of MALDI-imaging data sets. We discuss the importance of data preprocessing, which typically includes normalization, baseline removal and peak picking, and hightlight the importance of image denoising when visualizing IMS data.

Image Denoising Based on Compressed Sensing

Image Denoising Based on Compressed Sensing

A Research on SPOT5 Supermode Image Processing

Supermode is a completely original technique to improve the resolution of panchromatic images, there are some paper about this technique, but few regards the issue of Supermode image processing. we will focus on this issue in this paper. Based on a brief introduction about the supermode adopted in SPOT5 satellite, the sophisticated quincunx interpolation processing is considered, which can takes advantage of the regularity and structure inherent in interlaced data. Further more, some researches on denoising and deconvolution have been proposed and the main factors affect the PSF of SPOT5 imaging system are analyzed. Finally, we take some experiments on the SPOT5 supermode images processing and some satisfactory results are achieved.

An Adaptive Fuzzy-Wavelet Method of Signal Processing for Fiber Optic Gyroscopes

Fiber optical gyroscopes (FOGs) have been applied widely in navigation for mobile robots. The precision of outputs from FOGs is affected significantly by uncertain disturbances due to road conditions in practice. It is urgent to eliminate uncertain disturbances from outputs of sensors in harsh environments. This paper proposes an adaptive method of signal processing base on wavelet and fuzzy logic for outdoor mobile robots in different road condition. Property of disturbances in different outdoor road is taken into account to determine filtering thresholds in this research. Corresponding characteristics of real-time data were processed by a fuzzy logic inference system when mobile robots moved, and then they were de-noising by means of wavelet technique. It was proved that this proposed adaptive method can eliminate disturbances due to road irregularities and improve accuracy of surrogate data by experiments.

采用非局部主成分分析的极大似然估计图像去噪

A maximum likelihood estimation image denoising method is proposed using the non-local principle component analysis.Pixels with high similarity in both the gray level and the texture information are selected,and used to implement the maximum likelihood estimation.This kind of optimal restored method can overcome the drawback of the local image denoising method such as blurring edge,and improve the accuracy for restoring detail information in image using maximum likelihood estimation.Experiments using the proposed method,principal neighborhood dictionary non-local mean method and local maximum likelihood estimation method are implemented in images with different noise standard and different geometric complexity,and the performance of aboving three denoising methods are compared quantitatively and qualitatively.The results demonstrate that the proposed method can remove the noise effectively and preserve detail imformation of images compared with the currently used methods.

粗糙金属表面光条中心提取方法

To extract the laser stripe center of a rough metal surface accurately,an extraction method based on light cutting was presented.Firstly,the noise source of an image was analyzed by the image morphology,then the noise was converted into a granular shape while the continuous feature of an effective laser-stripe was still kept through an image enhancement processing.The noise and the laser-stripe were segmented by counting the connected areas of the image,and the denoised binary laser-stripe image was taken as a mask to be multiplied by the original image to recover the gray scale information of the laser-stripe image.Finally,the denoised laser-stripe center of the gray scale image was extracted with a barycenter method.Experimental results show that the laser-stripe centers extracted with the proposed method all locate in the laser-stripe region.The average error of the center extraction of a much-noised straight line laser-stripe by this method can reach 0.337 5 pixels.

基于小波变换的信号去噪在瑞利泰勒不稳定性定标实验中的应用

基于小波变换的信号去噪在瑞利泰勒不稳定性定标实验中的应用

A Wavelet-Based Mammographic Image Denoising and Enhancement with Homomorphic Filtering

Breast cancer continues to be a significant public health problem in the world. The diagnosing mammography method is the most effective technology for early detection of the breast cancer. However,...