基于期望最大化算法的音频取证中的篡改检测

音频取证中的插值检测是信号篡改检测的重要方面。因为信号的篡改经常伴随着重采样操作,而重采样后的插值信号会引入周期性信息。应用期望最大化（EM）算法能针对这种周期信息估计参数,从而检测出信号是否被篡改。为了使EM算法迭代效果更好,更适用于音频信号的插值检测问题,提出针对音频信号的特点,引入音频幅度直方图,排除短时静音和增加样本点数的方法。另外还提出了用频谱统计矩作为特征的方法,使统计分类稳定有效。最后通过音频取证中检测信号是否重采样的统计分类实验,表明整个检测流程能达到较高的准确率,并且在局部篡改实验中也同样有效。     

基于兴趣点的图像检索

在基于内容的图像检索中,往往使用颜色、纹理以及形状的全局特征来描述图像,然而全局特征不能描述图像的细节,丢失了图像的空间信息。文章利用兴趣点来灵活描述图像的局部信息,提取兴趣点周围的颜色矩作为局部特征,通过兴趣点的匹配和带权投票来进行相似度量,几何哈希技术的使用增强了兴趣点间的正确匹配。实验证明了这种方法的有效性,具有旋转、平移和部分的尺度不变性。     

Degradation of polycyclic aromatic hydrocarbons: Model simulation for bioavailability and biodegradation

A mathematical model for a one‐dimensional convective‐dispersive solute transport in a soil matrix is presented. The interplay of linear equilibrium sorption and first order degradation were incorporated into the formulation of the model. The model took into consideration the overall effects of the solid and liquid phase mass transfer resistances. Data from experiments were analysed using the temporal moments method (MOM) and CXTFIT curve‐fitting technique. The functional parameters; dispersion coefficient, first order degradation rate constant and the retardation factor were estimated using both first and second normalised moments. The solution to the model equations was achieved by the use of the backward finite difference scheme. Results obtained showed that naphthalene was more selectively degraded than pyrene and anthracene with a residual concentration of naphthalene 1.12E ? 5 mg/L, 1.48 mg/L; pyrene 3.11E ? 4 mg/L, 1.58 mg/L; and anthracene 7.67E ? 4 mg/L, 1.61 mg/L in the axial and radial directions, respectively. Concisely, the modeling results showed the occlusion of these compounds within the fissures and cavities of the soil particles, which renders them not readily bioavailable and thus inaccessible to microbial degradation.     

基于修正直方图不变矩的零水印算法

为了降低基于不变矩的零水印算法的漏检率和误检率,提出了一种基于修正直方图不变矩零水印算法.该算法计算原始图像的直方图不变矩,并利用其构建零水印系统,然后用参考图像和测试图像之间的相关系数来修正直方图不变矩,利用修正后的直方图不变矩来检测水印.实验结果表明,本算法水印检测过程简单,具有较低的漏检率和虚警率,与传统的基于直方图不变矩的零水印算法相比,在抵抗常规信号处理和几何攻击方面具有较好的鲁棒性.     

基于微分代数积分矩量法的聚并器超细颗粒聚团研究

研究聚并器内布朗聚团和湍流聚团引起超细颗粒聚团,特别考虑颗粒之间近程力（范德华引力、静电斥力）和颗粒与气体之间的流体力学作用力对颗粒聚团的影响。基于FLUENT软件UDF功能自定义聚团核,考虑颗粒之间近程力和流体力学作用力对聚并率的影响,引入碰撞效率α对聚团核进行修正,得到修正湍流聚并模型并将该模型与理想湍流聚并模型进行比较。应用群体平衡模型（population balance model,PBM）耦合CFD对颗粒聚团过程进行数值模拟,并采用微分代数积分矩量法（DAE-QMOM）求解群体平衡方程。结果表明：理想湍流聚并模型与实验结果误差为8.92%,而修正改进的湍流聚团模型与实验结果误差仅为3.35%,更加符合实际情况;微分代数积分矩量法具有较高的效率,而且误差较小,相比PD积分矩量法有明显的优势,稳定性也比较突出。     

多失效模式机械零件可靠性灵敏度估计

机械零件的失效往往存在多种失效模式,且这些失效模式对可靠性灵敏度均有影响,因此在进行可靠性灵敏度估计时,要充分考虑零件的各个潜在失效模式。提出一种多失效模式机械零件可靠性灵敏度分析的数值方法。运用随机摄动技术和四阶矩技术,求得各失效模式功能函数的前四阶矩及可靠性指标,解决了随机变量分布类型未知时,各失效模式的可靠度计算问题。根据机械零件失效模式较少的特点,将零件的潜在失效模式视为串联关系,则零件整体的失效即为失效模式的串联失效。结合灵敏度分析的梯度算法,推导出关于随机变量均值和方差的可靠性灵敏度的计算公式。数值算例表明所述方法可快速有效地估计多失效模式机械零件的可靠性灵敏度,且适用于非线性功能函数的情况。     

移动荷载识别的适用性研究

桥梁移动荷载识别技术的适用性是决定其能否直接应用于现场桥梁荷载在线识别的关键因素。基于矩量法和移动荷载识别理论,采用整域基函数表达桥面移动车载,提出了一种时域移动荷载识别改进方法(ITDM)。分别选用正交勒让德多项式和傅立叶级数作为基函数,形成两种改进算法(LITDM和FITDM),通过仿真及试验验证所提方法的正确性及可行性,并从基函数数量和运行时间两方面与时域法(TDM)进行对比研究。仿真及试验研究结果表明:LITDM、FITDM性能与TDM相近,但ITDM灵活性更强,其中以FITDM计算效率最高,适用性最强,LITDM次之,这对于桥梁移动荷载的现场在线准确识别非常有利。     

运动模糊不变量研究

图像的运动模糊是摄像机在曝光期间与场景间相对运动而造成的图像退化,但现有算法对运动模糊的恢复精度有限.本文分析了图像运动模糊的退化模型,提出了运动模糊不变量的概念,证明了图像零阶矩、第一个Hu不变矩和SIFT为运动模糊不变量,并给出了测试和应用.结果表明运动模糊不变量是存在的,而且可以用于不确定性运动模糊图像的匹配和识别领域.     

Solution of arbitrarily dimensional matrix equation in computational electromagnetics by fast lifting wavelet‐like transform

A new wavelet matrix transform (WMT), operated by lifting wavelet‐like transform (LWLT), is applied to the solution of matrix equations in computational electromagnetics. The method can speedup the WMT without allocating auxiliary memory for transform matrices and can be implemented with the absence of the fast Fourier transform. Furthermore, to handle the matrix equation of arbitrarily dimension, a new in‐space preprocessing technique based on LWLT is constructed to eliminate the limitation in matrix dimension. Complexity analysis and numerical simulation show the superiority of the proposed algorithm in saving CPU time. Numerical simulations for scattering analysis of differently shaped objects are considered to validate the effectiveness of the proposed method. In particular, due to its generality, the proposed preprocessing technique can be extended to other engineering areas and therefore can pave a broad way for the application of the WMT. Copyright © 2009 John Wiley & Sons, Ltd.     

Gaussian-Hermite moment-based depth estimation from single still image for stereo vision

Depth information of objects plays a significant role in image-based rendering. Traditional depth estimation techniques use different visual cues including the disparity, motion, geometry, and defocus of objects. This paper presents a novel approach of focus cue-based depth estimation for still images using the Gaussian-Hermite moments (GHMs) of local neighboring pixels. The GHMs are chosen due to their superior reconstruction ability and invariance properties to intensity and geometric distortions of objects as compared to other moments. Since depths of local neighboring pixels are significantly correlated, the Laplacian matting is employed to obtain final depth map from the moment-based focus map. Experiments are conducted on images of indoor and outdoor scenes having objects with varying natures of resolution, edge, occlusion, and blur contents. Experimental results reveal that the depth estimated from GHMs can provide anaglyph images with stereo quality better than that provided by existing methods using traditional visual cues.