广义梯度及其在盲信号分离中应用的研究

盲信号的研究是当今学术界研究的热点,它连接了神经网络、信号处理、信息论等多个领域[1-2].盲信号分离(B lind Signal Separation,简写为BSS)是盲信号的研究中一个广受关注的方向.在盲信号分离的研究中,1994年Comon[3]提出了独立分量分析(Independent Component Analysis,简写     

雷达信号的盲分离

盲信号处理已成为近年信号处理和神经网络热点领域.但主要用于语言信号处理,而应用于雷达信号处理并不多见.将盲源分离算法应用于雷达阵列接收信号处理,提出了一种新的盲源分离算法的性能评价标准—相关系数法.首先研究了雷达阵列接收模型,然后分析了雷达阵列接收信号的特性.由分析可知:当雷达目标和杂波源在空间上与雷达接收机的距离基本一致时,雷达阵列接收信号在雷达接收机上的混叠是瞬时混叠;同时雷达阵列接收信号均为超高斯信号.因此可采用盲源分离算法中的定点ICA算法来分离雷达阵列接收信号.仿真结果表明,分离出来的信号与源信号的相关系数均大于0.95,说明了定点ICA算法能雷达阵列接收信号,证明了理论分析的正确性和算法的有效性.     

Characterization of surfactant complex mixtures using Raman spectroscopy and signal extraction methods: Application to laundry detergent deformulation

This paper presents the analysis of surfactants in complex mixtures using Raman spectroscopy combined with signal extraction (SE) methods. Surfactants are the most important component in laundry detergents. Both their identification and quantification are required for quality control and regulation purposes. Several synthetic mixtures of four surfactants contained in an Ecolabel laundry detergent were prepared and analyzed by Raman spectroscopy. SE methods, Independent Component Analysis and Multivariate Curve Resolution, were then applied to spectral data for surfactant identification and quantification. The influence of several pre-processing treatments (normalization, baseline correction, scatter correction and smoothing) on SE performances were evaluated by experimental design. By using optimal pre-processing strategy, SE methods allowed satisfactorily both identifying and quantifying the four surfactants. When applied to the pre-processed Raman spectrum of the Ecolabel laundry detergent sample, SE models remained robust enough to predict the surfactant concentrations with sufficient precision for deformulation purpose. Comparatively, a supervised modeling technique (PLS regression) was very efficient to quantify the four surfactants in synthetic mixtures but appeared less effective than SE methods when applied to the Raman spectrum of the detergent sample. PLS seemed too sensitive to the other components contained in the laundry detergent while SE methods were more robust. The results obtained demonstrated the interest of SE methods in the context of deformulation.     

基于Gabor小波变换的ICA火灾图像纹理识别算法

针对火灾图像纹理识别问题,提出了基于Gabor小波变换的ICA火灾图像纹理识别算法,并根据火灾图像纹理识别特点进行了优化。首先用不同尺度和方向的Gabor滤波器对待识别图像滤波,得到其特征图像,然后将特征图像转化成特征向量作为ICA的输入,得到基矢量子空间,再将测试图像经过Gabor滤波器的特征向量投影到ICA子空间中得到系数向量作为目标识别特征,最后用支持向量机进行识别。通过与Gabor滤波器法和ICA方法的对比实验,表明该算法可以在火灾纹理图像的识别率上比传统方法提高5%以上,为火灾图像识别提供了一种新思路。     

Detecting functional connectivity in the resting brain: a comparison between ICA and CCA

Independent component analysis (ICA) and cross-correlation analysis (CCA) are general tools for detecting resting-state functional connectivity. In this study, we jointly evaluated these two approaches based on simulated data and in vivo functional magnetic resonance imaging data acquired from 10 resting healthy subjects. The influence of the number of independent components (maps) on the results of ICA was investigated. The influence of the selection of the seeds on the results of CCA was also examined. Our results reveal that significant differences between these two approaches exist. The performance of ICA is superior as compared with that of CCA; in addition, the performance of ICA is not significantly affected by structured noise over a relatively large range. The results of ICA could be affected by the number of independent components if this number is too small, however. Converting the spatially independent maps of ICA into z maps for thresholding tends to overestimate the false-positive rate. However, the overestimation is not very severe and may be acceptable in most cases. The results of CCA are dependent on seeds location. Seeds selected based on different criteria will significantly affect connectivity maps.     

基于可见光源与独立成分分析法的实时人眼检测系统

提出一种基于特征与外貌混合检测确定人眼区域的实时人眼检测方法. 首先,依据可见光源在人眼角膜上反射形成耀点特性,通过图像处理算法提取潜在耀点位置,利用人眼几何特征的确定可能人眼候补区域;然后,提取人眼数据库中具有不同外貌特征的200幅人眼图像,采用FastICA算法估计出提取人眼图像的有效成分分析(ICA)基向量;最后,通过计算人眼候选区域在基向量上投影角度判断出左、右人眼区域准确位置. 实验结果表明,在人脸面部旋转、佩戴眼镜、大范围头部运动和不同光照强度下,实时人眼检测具有较高的检测正确率和较好的鲁棒性.      

基于独立成分分析的高光谱图像异常检测

针对高光谱图像中背景及目标先验知识未知条件下的异常目标检测问题,提出了一种基于独立成分分析(ICA)的异常探测算法.首先估计原始数据的虚拟维(VD)以确定要分离的独立成分个数,在此基础上进行快速独立成分分析(FastICA),然后基于平均局部奇异度选择含异常信息较多的独立成分,最后使用丰度量化算法得到异常目标的丰度图像...     

Independent components analysis coupled with 3D-front-face fluorescence spectroscopy to study the interaction between plastic food packaging and olive oil

Olive oil is one of the most valued sources of fats in the Mediterranean diet. Its storage was generally done using glass or metallic packaging materials. Nowadays, plastic packaging has gained worldwide spread for the storage of olive oil. However, plastics are not inert and interaction phenomena may occur between packaging materials and olive oil. In this study, extra virgin olive oil samples were submitted to accelerated interaction conditions, in contact with polypropylene (PP) and polylactide (PLA) plastic packaging materials. 3D-front-face fluorescence spectroscopy, being a simple, fast and non destructive analytical technique, was used to study this interaction. Independent components analysis (ICA) was used to analyze raw 3D-front-face fluorescence spectra of olive oil. ICA was able to highlight a probable effect of a migration of substances with antioxidant activity. The signals extracted by ICA corresponded to natural olive oil fluorophores (tocopherols and polyphenols) as well as newly formed ones which were tentatively identified as fluorescent oxidation products. Based on the extracted fluorescent signals, olive oil in contact with plastics had slower aging rates in comparison with reference oils. Peroxide and free acidity values validated the results obtained by ICA, related to olive oil oxidation rates. Sorbed olive oil in plastic was also quantified given that this sorption could induce a swelling of the polymer thus promoting migration.     

基于独立分量分析的跳频通信抗梳状阻塞干扰方法

针对梳状阻塞干扰对跳频通信造成严重影响的问题,根据跳频信号与梳状干扰的统计独立特性,将独立分量分析引入跳频通信抗干扰之中,提出了基于独立分量分析的跳频通信抗梳状干扰方法。围绕输出信号的统计独立性构建对照函数,利用对照函数引导分离矩阵迭代,进而将分离矩阵作用于接收混合信号,实现跳频信号与干扰信号的有效分离。仿真结果表明：提出的方法可有效提高跳频通信对抗梳状阻塞干扰能力;当误码率为10-3时,提出的方法可使跳频通信的抗干扰能力提升约8 dB。