基于子带ICA的时频图像处理方法研究及其在故障诊断中的应用

提出了一种改进独立分量分析(ICA)应用于时频图像的盲源分离问题。由于相似时频图像之间存在潜在的相关性,传统的ICA对于具有相关成分的时频图像盲源分离中效果比较差,利用互信息和峭度研究了图像子带之间的相关性和本身的非高斯性,选定特定的子带进行ICA分析。通过仿真时频图像的分离试验,说明此方法分离效果明显优于ICA分离效果,并将该方法应用于转子试验台的基座松动,不对中故障信号复合故障的时频图像中,成功获取了各自故障的时频图像,从而可以获得各自的故障特征信息。     

基于独立分量分析的高速微铣削力混合信号噪声分离方法

为得到高速微铣削力的真实信号,并且准确识别各激励源,对微铣削力混合信号分离和识别技术进行研究。首先对铣削力混合信号矩阵进行预处理,利用对预处理结果的独立成分分析(ICA)分离得到独立源信号矩阵,再通过快速傅里叶变换(FFT)得到独立源信号的频谱,最后分析并结合实验工况识别出微铣削力信号、机械噪声信号和环境噪声信号。实验结果表明:该方法具有同时分离非高斯性的机械噪声信号和高斯性的环境噪声信号的优点,可以弥补传统方法只能抑制高斯噪声信号的不足。     

Application of independent component analysis with mixture density model to localize brain alpha activity in fMRI and EEG

Independent component analysis (ICA) is an approach to solve the blind source separation problem. In the original and extended versions of ICA, nonlinearity functions are fixed to have specific density forms such as super‐Gaussian or sub‐Gaussian, thereby limiting their performance when sources with different classes of densities are mixed in multichannel data. In this article, we have incorporated a mixture density model such that no assumption about source density would be required. We show that this leads to better source separation due to increased flexibility in handling source‐ densities with flexible parametric nonlinearity. The algorithm was validated through simulation studies and its performance was compared to other versions of ICA. The modified mixture density ICA was then applied to functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) data to localize independent sources of alpha activity in the human brain. A good spatial correlation was found in the spatial distribution of alpha sources derived independently from fMRI and EEG, suggesting that spontaneous alpha rhythm can be imaged by fMRI using ICA without concurrent acquisition of EEG. © 2004 Wiley Periodicals, Inc. Int J Imaging Syst Technol 14, 170–180, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.20021     

独立分量分析及其在故障诊断中的应用

独立分量分析是盲源分离的一种新方法，其处理的对象是相互统计独立的信号源经线性组合而产生的一组混合信号，最终目的是从混合信号中分离出各独立的信号分量。本简要介绍了独立分量分析的基本思想及算法，并对现场采集到的多组振动信号进行了分析，结果表明，独立分量分析在对混合信号进行盲分离方面具有很强的能力，从而为机械设备状态监测与故障诊断提供了一种行之有效的信号预处理的新方法。     

Cutting force denoising in micro-milling tool condition monitoring

An independent component analysis (ICA) algorithm for cutting force denoising was applied in micro-milling tool condition monitoring. In micro-milling, the comparatively small cutting force signal is prone to contamination by relatively large noise, and as a result it is important to denoise the force signal before further processing it. However, the traditional denoising methods, based on Gaussian noise assumption, lose here because the noise is identified as containing a high non-Gaussian component in the experiment. ICA was recently developed to deal with the blind source separation (BSS) problem. It solves the BSS problem by measuring the non-Gaussianity of the signal and it is particularly effective in the separation of non-Gaussian signals. This approach employs fixed-point ICA (FastICA), assuming the noises are sources and the force signal is an instantaneous mixture of sources and by treating the signal denoising process as a BSS. The results are illustrated both in time and frequency domains. The FastICA denoising performances are compared with the popular wavelet thresholding. The results show that FastICA performs better than wavelet. Theoretical discussion of the nature of ICA and wavelet thresholding supports the results: ICA separates both Gaussian and non-Gaussian noise sources, while wavelet only suppresses Gaussian noise.     

Approach to blind image deconvolution by multiscale subband decomposition and independent component analysis

A single-frame multichannel blind image deconvolution technique has been formulated recently as a blind source separation problem solved by independent component analysis (ICA). The attractive feature of this approach is that neither origin nor size of the spatially invariant blurring kernel has to be known. To enhance the statistical independence among the hidden variables, we employ multiscale analysis implemented by wavelet packets and use mutual information to locate a subband with the least dependent components, where the basis matrix is learned by means of standard ICA. We show that the proposed algorithm is capable of performing blind deconvolution of nonstationary signals that are not independent and identically distributed processes. The image poses these properties. The algorithm is tested on experimental data and compared with state-of-the-art single-frame blind image deconvolution algorithms. Our good experimental results demonstrate the viability of the proposed concept.     

Independent Component Analysis–Based Feature Extraction Technique for Defect Classification Applied for Pulsed Eddy Current NDE

With the development of nondestructive detection, the emerging testing techniques provide new challenges to signal analysis and interpretation approach applied to the inspection evaluation. Some researchers have developed the methods that focus on feature analysis of detected signals. This article presents a new feature analysis by the Independent Component Analysis (ICA) approach to evaluate the defects tested by the pulsed eddy current (PEC) technique. ICA is a high-order statistics technique used to separate multi-unknown sources, which has been successfully applied to facial image identification and separation of the components of 1D signal. In this article, the ICA approach is utilized to project the response signals of various defects into the independent components (ICs) feature subspace by signal representation model. Dependent on the selected ICs, each defect is represented by different projected coefficients, which are proposed to discriminate and classify the defects that belong to three categories. The improved ICA model is proposed to improve the classification of two similar categories of single defects: metal loss and subsurface defects. The evaluation using the series of experimental data has validated the classification of single defects and the defects with lift-off effect by our ICA approach. The comparison with Principal component analysis (PCA)–based approach further verified the better performance of the ICA-based model.     

相关机械振源的盲源分离方法

相关源不满足独立分量分析关于源的统计独立性假设,标准的独立分量分析方法无法准确分离相关机械振源信号。在相关振源信号的部分频带满足统计独立的假设前提下,提出了一种基于小波包分解的相关机械源盲源分离方法。该方法将观测信号用小波包分解成子带观测信号,根据互信息标准选择相关性较小的若干子带观测信号重构观测信号。通过重构的观测信号的独立分量分析估计分离矩阵,然后用该矩阵分离原始观测信号从而实现相关机械振源信号的分离。仿真试验验证了该方法的有效性。     

Source extraction from functional magnetic resonance imaging data using ICA based on fourth-order and exponential contrast functions

Independent component analysis (ICA) is one of the well-known statistical techniques used for blind source separation. It is also used for the extraction of sources from functional magnetic resonance imaging (fMRI) data. Benchmark for different ICA algorithms is speed and accuracy. In this article, we will be focusing on two simple contrast functions along with matrix-based updating rules. Fixed-point iteration is used for optimization of the contrast functions. Application of matrix-based weight updating makes the process converge rapidly. Validity of the algorithms is tested by comparing the speed and accuracy on simulated and actual fMRI data with other conventional ICA approaches.     

基于独立分量分析的形状识别

物体的形状识别是模式识别的重要方向之一,广泛应用于图像分析、机器视觉和目标识别等领域。在介绍利用信号的高阶统计信息的独立分量分析方法基础上,提出了基于独立分量分析的形状识别方法。利用独立分量分析算法提取出图像的独立基,根据待识别图像在独立基上投影系数的差别进行分类识别。仿真实验结果表明,该方法对于形状识别有较高的识别率。     

Geometric transformation based independent component analysis for mixed image separation

Abstract

An independent component analysis (ICA) method for image separation by geometric transformation of a scatter diagram is proposed. Geometric transformation and normalization are used to project mixed image signals to independent component space. This method includes four procedures: data correction, whitening, geometric rotation, and slant compensation. Several synthetic mixed image and real applications are used to evaluate the performance of the proposed method. From experimental results, mixed images are separated accurately by the proposed method.     

基于独立分量分析的爆破振动信号分离仿真试验

爆破振动信号分析是研究爆破振动特性的重要手段,而实测爆破振动信号往往是多个振源信号的叠加信号,分离出每个振源或特定振源的振动信号对于其振动特性的研究具有重要意义。选取实测的爆破振动信号,进行线性混合得到混合信号,利用独立分量分析(ICA)对混合信号进行分离,获取了与源信号波形基本一致的分离信号,仿真试验表明ICA适合爆破振动信号的分离。     

基于粒子群算法优化的独立分量分析算法

通过两组模拟信号对三种主流独立分量分析算法-JADE、FastICA、扩展Infomax算法的性能进行了对比分析,结果表明三种算法均无法完全分离超高斯源与亚高斯源形成的混合信号,FastICA算法对能量强弱差别大的混合信号失效。基于这一现象,提出了一种新的独立分量分析算法,以粒子群算法为优化工具,以分离矩阵为优化变量,最小化分离信号联合概率与边缘概率乘积的差值,并给出了具体的计算流程。仿真实验结果表明,该算法的性能显著优于上述三种独立分量分析算法。同时,新提出算法实施过程中不需要任何先验知识,相比其他三种ICA算法,更适合解决工程实际问题。最后,将该算法应用于对滚动轴承实验台实测信号的处理,通过对分离信号的分析实现了对滚动轴承故障类型的准确识别,进一步证明了算法的有效性。     

应用ICA滤波器技术提取图像纹理特征

针对纹理图像分类问题,本文提出了一种应用ICA滤波器技术提取图像纹理特征的方法.该方法首先从训练图像集中随机抽取图像块作为观测信号,应用ICA技术,提取滤波器组.然后根据训练样本图像对滤波器组的响应值来评估和选择滤波器组,达到降维的目的.最后利用滤波器组对测试图像进行滤波,得到该图像的滤波响应结果,从该响应结果中得到最大响应滤波器编号,提取其直方图作为图像的全局特征和局部特征.对Brodatz纹理图像集中108个纹理类别进行了分类实验,结果表明,与MPEG-7纹理描述子相比,该图像特征对纹理图像具有更好的分类效果.     

基于盲源分离技术的一种信号处理方法研究

文章利用盲源分离技术试图从给定的一组混合观察数据中恢复未知的独立信源。然后使用Matlab中的时频工具箱,根据分离出的信号进行时频分析,提取信号中所包含的特征信息,达到能够识别出期望信号,并能提取其余干扰信号的特征。     

一种用于语音增强的频域盲信号分离算法

研究了在未知声源信息和传声器空间位置的情况下,利用盲信号分离的方法实现语音增强。通过把基于信息论的信息最大化算法推广到频域,使得时域的卷积混合问题转变为频域的瞬时混合问题,进而就可以在每个频段分别进行独立分量分析,分离效果有明显改进,算法收敛性也得到提高。为了克服在频域中实现盲分离时所固有的位序不确定性和比例缩放问题对分离性能的严重影响,采用聚类的方法对每个频率段的分离结果进行排序。对真实环境中录制的语音、音乐混合信号和语音、语音混合信号进行了计算机仿真,分离之后使语音的信噪比提高了10-15dB,很好地实现了语音增强的目的。     

振动噪声信号分离优化算法研究

机械振动信号中混杂的背景噪声对信号的分析处理以及振动结构的建模有很大影响,在独立分量分析（ICA）理论研究的基础上,以信息论中的最小化互信息准则作为ICA的判据,提出了一种基于负熵的FastICA共轭梯度快速算法。该算法计算效率高,收敛性能好;适合对初始点不敏感和对鲁棒性有要求的信号处理。实验结果表明了对振动信号的背景噪声分离具有较好的效果。     

Multivariate statistical analysis of Raman images of a pharmaceutical tablet

This paper describes the application of principal component analysis (PCA) and independent component analysis (ICA) to identify the reference spectra of a pharmaceutical tablet's constituent compounds from Raman spectroscopic data. The analysis shows, first with a simulated data set and then with data collected from a pharmaceutical tablet, that both PCA and ICA are able to identify most of the features present in the reference spectra of the constituent compounds. However, the results suggest that the ICA method may be more appropriate when attempting to identify unknown reference spectra from a sample. The resulting PCA and ICA models are subsequently used to estimate the relative concentrations of the constituent compounds and to produce spatial distribution images of the analyzed tablet. These images provide a visual representation of the spatial distribution of the constituent compounds throughout the tablet. Images associated with the ICA scores are found to be more informative and not as affected by measurement noise as the PCA based score images. The paper concludes with a discussion of the future work that needs to be undertaken for ICA to gain wider acceptance in the applied spectroscopy community.     

Medical image registration‐based retrieval using distance metrics

In this article, registration and retrieval are carried out separately for medical images and then registration‐based retrieval is performed. It is aimed to provide a more thorough insight on the use of registration, retrieval, and registration‐based retrieval algorithm for medical images. The purpose of this work is to deal these techniques with anatomical imaging modalities for clinical diagnosis, treatment, intervention, and surgical planning in a more effective manner. Two steps are implemented. In the first step, the affine transformation‐based registration for medical image is processed. The second step is the retrieval of medical images processed by using seven distance metrics such as euclidean, manhattan, mahalanobis, canberra, bray‐curtis, squared chord, chi‐squared, and also by using the features like mean, standard deviation, skewness, energy, and entropy. Now images registered by affine transformation are applied for retrieval. In this work, both registration and retrieval techniques in medical domain share some common image processing steps and required to be integrated in a larger system to complement each other. Experimental results, it is evident that euclidean and manhattan produces 100% precision and 35% recall found to have higher performance in retrieval. From the four anatomical modalities considered (brain, chest, liver, and limbs) brain image has better registration. It is also found that though the registration of images changes the orientation, for better performance of images in clinical evaluation it does not widely affect the retrieval performance. In the medical domain the ultimate aim of this work is to provide diagnostic support to physicians and radiologists by displaying relevant past cases, along with proven pathologies as ground truth from experimental results. © 2013 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 23, 360–371, 2013     

基于变分贝叶斯独立分量分析的故障源盲分离

提出一种基于变分贝叶斯独立分量分析的故障源盲分离方法,该方法可直接对噪声干扰的机械源信号进行有效分离,即不需要将未知噪声看成一种独立源,也不需要进行消噪预处理。并将该方法与传统的机械源分离方法进行对比实验,实验结果表明该方法是非常有效的。