基于YCbCr颜色空间背景建模与手势阴影消除

为了提高动态手势检测的精确度,本文将基于YCbCr颜色空间的混合高斯背景建模应用于动态手势识别中,并且提出手势阴影消除的有效算法。首先,对待检测视频帧通过抠图抠出手势图像,在YCb'Cr'颜色空间进行椭圆拟合,统计建立椭圆肤色模型,继而在YCbCr颜色空间进行混合高斯背景建模检测出动态手势,点乘原图像得到含有阴影的RGB手势图像,对检测出的含有阴影的手势图像利用已建立的椭圆肤色模型进行阴影消除,最后将手势图像连成视频序列。实验结果表明,该算法在复杂背景下进行动态手势的检测率可达91.4%,高出传统方法10%左右,能够满足动态手势检测基本要求,且具有较高的实用价值。     

Background subtraction based level sets for human segmentation in thermal infrared surveillance systems

Based on the technique of background subtraction, two level set based active contour models (LSACs) named as RT-BSLSAC and EA-BSLSAC are proposed for human segmentation in thermal infrared surveillance systems. The energy functional of RT-BSLSAC is initially formulated with the spatial–temporal information extracted from the background-subtracted images that correspond to the current frame and its adjacent frames. Then, minimization of such functional is conducted by a real-time numeric scheme evolving a binary level set function (BLSF). When the BLSF converges, the moving humans in current frame are detected with relatively complete interiors and enclosed, smooth contours. EA-BSLSAC makes two improvements to RT-BSLSAC. First, the formulation of energy functional not only depends on spatial–temporal information but also the boundary information resulting from an edge detector. Second, the functional is minimized by a convex numeric scheme featured by initialization-invariance. As a result, EA-BSLSAC presents higher segmentation accuracy but at more computational cost in comparison with RT-BSLSAC. Experimental results from segmenting the real-world infrared surveillance clips validate the advantages of the proposed methods in accuracy, efficiency, and the coordination with other algorithmic components of an infrared surveillance system due to the cancellation of post-processing meaning to reach complete human interiors and exact silhouettes.     

Distributed analysis of simultaneous EEG-fMRI time-series: modeling and interpretation issues

Functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) represent brain activity in terms of a reliable anatomical localization and a detailed temporal evolution of neural signals. Simultaneous EEG-fMRI recordings offer the possibility to greatly enrich the significance and the interpretation of the single modality results because the same neural processes are observed from the same brain at the same time. Nonetheless, the different physical nature of the measured signals by the two techniques renders the coupling not always straightforward, especially in cognitive experiments where spatially localized and distributed effects coexist and evolve temporally at different temporal scales.     

Significance analysis and statistical mechanics: an application to clustering

This Letter addresses the statistical significance of structures in random data: given a set of vectors and a measure of mutual similarity, how likely is it that a subset of these vectors forms a cluster with enhanced similarity among its elements? The computation of this cluster p value for randomly distributed vectors is mapped onto a well-defined problem of statistical mechanics. We solve this problem analytically, establishing a connection between the physics of quenched disorder and multiple-testing statistics in clustering and related problems. In an application to gene expression data, we find a remarkable link between the statistical significance of a cluster and the functional relationships between its genes.     

Speech dereverberation method based on spectral subtraction and spectral line enhancement

Speech signals recorded with a distant microphone usually are interfered by the spatial reverberation in the room, which severely degrades the clarity and intelligibility of speech. A speech dereverberation method based on spectral subtraction and spectral line enhancement is proposed in this paper. Following the generalized statistical reverberation model, the power spectrum of late reverberation is estimated and removed from the reverberation speech by the spectral subtraction method. Then, according to the human auditory model, a spectral line enhancement technique based on adaptive post-filtering is adopted to further eliminate the reverberant components between adjacent speech formants. The proposed method can effectively suppress the spatial reverberation and improve the auditory perception of speech. The subjective and objective evaluation results reveal that the perceptual quality of speech is greatly improved by the proposed method.     

基于在线小波支持向量回归的混沌时间序列预测

混沌时间序列预测是非线性动力学研究中一个十分重要的问题，支持向量回归方法为其提供了一种有效的解决思路.通过分析新样本加入训练集后支持向量集的变化情况，建立了一种混沌时间序列预测的支持向量回归算法，具备了在线学习的特点.同时，针对混沌信号提出了一种满足小波框架的小波核函数，它不但能以较高的精度逼近任意函数，而且适合于混沌信号的局部分析，提高了支持向量回归的泛化能力.最后就Mackey-Glass混沌时间序列在线预测问题进行了大量仿真.结果表明，本文算法与现有的算法相比具有训练时间短、预测精度高等特点，有一定 关键词： 混沌时间序列 支持向量回归 在线学习 小波核     

Energy gap,clustering, and the Goldstone theorem in statistical mechanics

We prove a Goldstone-type theorem for a wide class of lattice and continuum quantum systems, both for the ground state and at nonzero temperature. For the ground state (T=0) spontaneous breakdown of a continuous symmetry implies no energy gap. For nonzero temperature, spontaneous symmetry breakdown implies slow clustering (noL ¹ clustering). The methods apply also to nonzero-temperature classical systems.Partial financial support by Fundação de Amparo à Pesquisa do Estado de São Paulo.Partial financial support by CNPq.     

Improving statistical keyword detection in short texts: Entropic and clustering approaches

In the last years, two successful approaches have been introduced to tackle the problem of statistical keyword detection in a text without the use of external information: (i) The entropic approach, where Shannon’s entropy of information is used to quantify the information content of the sequence of occurrences of each word in the text; and (ii) The clustering approach, which links the heterogeneity of the spatial distribution of a word in the text (clustering) with its relevance. In this paper, first we present some modifications to both techniques which improve their results. Then, we propose new metrics to evaluate the performance of keyword detectors based specifically on the needs of a typical user, and we employ them to find out which approach performs better. Although both approaches work well in long texts, we obtain in general that measures based on word-clustering perform at least as well as the entropic measure, which needs a convenient partition of the text to be applied, such as chapters of a book. In the latter approach we also show that the partition of the text chosen affects strongly its results. Finally, we focus on short texts, a case of high practical importance, such as short reports, web pages, scientific articles, etc. We show that the performance of word-clustering measures is also good in generic short texts since these measures are able to discriminate better the degree of relevance of low frequency words than the entropic approach.     

Background subtraction: I. General behaviour of Tougaard-style backgrounds in AES and XPS

An analysis is made of Tougaard-style backgrounds for homogeneous materials to show how parameters interact together and comprise the general inelastic scattering background in electron spectroscopy. It is shown that insight is usefully gained by rewriting the Tougaard universal loss function in terms of two parameters, the characteristic energy, E₁, of the exponential decay observed in multiple self-convolutes of the Tougaard universal single loss function, and the centroid energy of the single loss function. Close fits to the measured background may be made over a wide energy range for only one value of E₁ which defines a unique relation between Tougaard’s B and C values but does not give the unique values themselves. For the single value of E₁, the centroid energy may be varied such that the background subtracted spectra range from those shown by Tougaard, where the intrinsic shake-up and losses may constitute two thirds of the peak intensity, to spectra similar to those of Jo in which all of the intrinsic losses are removed. Studies of Al X-ray excited Sc, Cu, Sm and Au photoelectron spectra, each of which has an extensive range of peaks, show that the relative intensities of the peaks are unaffected by the choice of the value of even though the absolute peak areas may change by a factor of 3.     

Multiparticle entangled quantum states and modeling of thermodynamic statistical distributions

Thermodynamic equilibrium of a system is considered as a consequence of quantum entanglement of the vacuum state of the system. An explicit mathematical model of multiparticle entangled pure quantum states is developed and analyzed. Within the framework of this model, the measurement process gives rise to probability distributions that exactly correspond to thermal equilibrium of the system in a thermostat.     

High-order fuzzy time-series based on multi-period adaptation model for forecasting stock markets

Stock investors usually make their short-term investment decisions according to recent stock information such as the late market news, technical analysis reports, and price fluctuations. To reflect these short-term factors which impact stock price, this paper proposes a comprehensive fuzzy time-series, which factors linear relationships between recent periods of stock prices and fuzzy logical relationships (nonlinear relationships) mined from time-series into forecasting processes. In empirical analysis, the TAIEX (Taiwan Stock Exchange Capitalization Weighted Stock Index) and HSI (Heng Seng Index) are employed as experimental datasets, and four recent fuzzy time-series models, Chen’s (1996), Yu’s (2005), Cheng’s (2006) and Chen’s (2007), are used as comparison models. Besides, to compare with conventional statistic method, the method of least squares is utilized to estimate the auto-regressive models of the testing periods within the databases. From analysis results, the performance comparisons indicate that the multi-period adaptation model, proposed in this paper, can effectively improve the forecasting performance of conventional fuzzy time-series models which only factor fuzzy logical relationships in forecasting processes. From the empirical study, the traditional statistic method and the proposed model both reveal that stock price patterns in the Taiwan stock and Hong Kong stock markets are short-term.     

Consistency in statistical moments as a test for bubble cloud clustering

Frequency dependent measurements of attenuation and/or sound speed through clouds of gas bubbles in liquids are often inverted to find the bubble size distribution and the void fraction of gas. The inversions are often done using an effective medium theory as a forward model under the assumption that the bubble positions are Poisson distributed (i.e., statistically independent). Under circumstances in which single scattering does not adequately describe the pressure field, the assumption of independence in position can yield large errors when clustering is present, leading to errors in the inverted bubble size distribution. It is difficult, however, to determine the existence of clustering in bubble clouds without the use of specialized acoustic or optical imaging equipment. A method is described here in which the existence of bubble clustering can be identified by examining the consistency between the first two statistical moments of multiple frequency acoustic measurements.     

Background subtraction in electron spectroscopy by use of reflection electron energy loss spectra

The use of reflected electron energy loss spectra (REELS) in deconvoluting the inelastic background signal from XPS and AES spectra from homogeneous samples is studied. It is demonstrated that under certain assumptions, the cross section for inelastic electron scattering can be extracted from a REELS spectrum. This cross section is applied to deconvolute an experimental XPS spectrum of aluminium. The method, its limitations and its relation to other methods are discussed.     

Adaptive contour-based statistical background subtraction method for moving target detection in infrared video sequences

A robust contour-based statistical background subtraction method for detection of non-uniform thermal targets in infrared imagery is presented. The foremost step of the method comprises of generation of background frame using statistical information of an initial set of frames not containing any targets. The generated background frame is made adaptive by continuously updating the background using the motion information of the scene. The background subtraction method followed by a clutter rejection stage ensure the detection of foreground objects. The next step comprises of detection of contours and distinguishing the target boundaries from the noisy background. This is achieved by using the Canny edge detector that extracts the contours followed by a k-means clustering approach to differentiate the object contour from the background contours. The post processing step comprises of morphological edge linking approach to close any broken contours and finally flood fill is performed to generate the silhouettes of moving targets. This method is validated on infrared video data consisting of a variety of moving targets. Experimental results demonstrate a high detection rate with minimal false alarms establishing the robustness of the proposed method.     

基于灰度聚类算法的红外图像增强研究

根据红外灰度图像的特点，提出了一种基于K-均值聚类的图像增强的新算法。该算法首先根据具体图像确定K值，其次对红外图像的辐射温度数据进行统计学习，把不同温度值按升序排列，然后按等差原则选取温度值作为初始聚类中心，再依据初始聚类中心采用K-均值聚类算法对温度进行聚类，最后由聚类结果对图像进行自适应增强。通过对红外灰度图像进行实验，得到了满意的结果: 对比直方图均衡，具有更丰富的图像细节信息和层次感，视觉效果更好。     

改进的K均值聚类红外目标检测方法

利用图像方差能很好地反映目标边缘信息的特点,提出一种基于方差的K均值聚类红外目标检测算法。利用形态学方法对红外图像进行预处理,运用相应的模板计算得到红外图像的方差图像,利用K均值聚类算法对方差图像进行聚类,从而分离出目标类别和背景类别。实验表明,该算法提取的红外图像中目标信息的兰德指数最高,说明该算法能有效地提取红外图像中目标信息,从而达到目标检测的目的。     

基于深度聚类的目标细粒度分类方法北大核心CSCD

为提高光电系统对弱小目标的识别和分类能力,降低算法对硬件平台和数据的依赖,提出一种无监督分类方法−基于目标深度特征聚类的细粒度分类方法。该方法通过轮廓、颜色、对比度等浅层特征提取提示目标,经超分辨处理后,利用卷积神经网络对目标的深层特征进行编码,进一步采用基于注意机制的主成分分析方法进行降维生成表征矩阵,最后利用聚类的方式实现目标细粒度分类。实验验证了基于不同神经网络的深度聚类方法在不同数据集上的分类性能,其中采用ResNet-34聚类方法在CIFAR-10测试集上细粒度分类性能达92.71%,结果表明,基于深度聚类的目标细粒度方法能够取得与强监督学习方法相当的目标分类效果。此外,还可以根据不同簇数和聚类等级的选择实现不同细粒度的分类效果。     

Self-organizing map clustering based on continuous multiresolution entropy

The detection of changes in the parameter values of a nonlinear dynamic system is a branch of study with multiple applications. In this paper, we explore a variant of an automatic detector and clustering of slight parameter variations in nonlinear dynamic systems proposed by Torres et al. [Automatic detection of slight changes in nonlinear dynamical systems using multiresolution entropy tools, Int. J. Bifurc. Chaos 11(4) (2001) 967–981]. The new method takes the advantages of the continuous multiresolution entropy to localize slight changes in the parameters, and uses self-organizing maps to quantify and cluster these changes. We discuss the performance of this method while applied to automatic segmentation of natural and synthetic diphthongs in the presence of additive noise. Our results show the potentiality of the proposed method.     

基于噪声谱结构特性的谱减法

提出了基于噪声谱结构特性的谱减法,在不增加语音失真的情况下,抑制传统谱减法的“音乐噪声”。首先,依据噪声谱结构特性在频带间自适应平滑周期图,减小谱估计方差的同时,避免噪声非连续谱的能量泄露;其次,依据噪声谱的结构特性,对增益函数进行自适应调整以更有效的抑制有调噪声。测试结果表明,不论对宽带噪声还是对窄带噪声,本文算法在信噪比提高和噪声抑制量等客观评价指标上都明显优于传统谱减法。非正式主观测听进一步验证了本文算法的有效性。     

A background subtraction approach based on complex wavelet transforms in EDXRF

Subtracting the background accurately is one of the most important issues in energy‐dispersive X‐ray fluorescence (EDXRF) spectra processing. This paper presents a novel approach to perform background subtraction based on dual‐tree complex wavelet transform. Compared with real wavelet transform, the proposed method has some attractive properties, including a smooth, nonoscillating, and nearly shift‐invariant magnitude with a simple near‐linear phase encoding of signal shifts. Therefore, it outperforms real wavelet transform to decompose background into low‐frequency components. The effectiveness of the proposed approach is demonstrated via two simulated spectra with different kinds of backgrounds and one measured spectrum from an energy‐dispersive X‐ray spectrometer. Both simulated and experimental results prove that the proposed approach can subtract background in EDXRF spectra effectively. Copyright © 2014 John Wiley & Sons, Ltd.