基于优化辅助粒子算法的红外目标跟踪研究

针对粒子算法存在的问题,提出了辅助粒子算法.该算法在重采样算法基础上,引进辅助变量,对粒子的权2次计算,可使粒子权值比重采样的粒子权值变化更稳定,最后给出了红外目标模型和均方根误差函数.仿真结果表明该算法对运动目标跟踪的均值和方差上均优于标准粒子滤波、重采样粒子滤波,且提高了计算效率.     

基于辅助粒子滤波算法的红外目标跟踪

针对红外目标在跟踪中计算复杂的问题,构建辅助粒子滤波算法。利用贝叶斯重要性采样算法,在权值大的粒子基础上引入辅助粒子变量,然后重新定义重要采样分布函数,防止重采样后粒子概率密度变化。两次加权计算,使粒子权值比仅用重采样的粒子权值变化更稳定,采样点最接近真实状态;同时不同权值粒子的概率阈值可作为粒子滤波是否完成的判断准则。在二维平面构造红外运动目标模型中,系统为零均值高斯白噪声。仿真数据表明:该算法在x,y方向的均方误差、画面处理时间、RM SE性能上优于粒子滤波算法和重采样粒子滤波算法。     

基于多速率交互式多模型的快速光电跟踪算法

针对交互式多模型粒子滤波算法运算量大的问题,提出了一种基于多速率跟踪思想的交互式多模型算法.该算法根据各模型假定的机动性,采用不同的数据更新速率,实现了模式空间和测量空间的混合滤波.同时,多模型综合选配了不同的滤波算法,其中,弱机动模型匹配卡尔曼滤波器,强机动模型匹配粒子滤波器.仿真结果表明,与传统的交互式多模型粒子滤波算法相比,本文算法在保证滤波精度的基础上,具有较低的计算复杂度,降低了约38.9%,能够有效地改善光电目标跟踪系统的可靠性和实时性.     

一种基于积分直方图的粒子滤波跟踪方法

用积分直方图技术提出了一种低复杂度的粒子滤波跟踪方法和基于积分颜色直方图的观测似然模型.采用积分方向直方图建立一个检测响应图,用其上的观测信息构建建议分布函数,在状态空间中似然概率较大的子空间中进行粒子采样.仿真结果表明,在采用大量粒子跟踪大目标时,提出的跟踪方法的计算复杂度明显低于直方图直接提取的粒子滤波方法,而且在...     

一种基于积分直方图的粒子滤波跟踪方法

用积分直方图技术提出了一种低复杂度的粒子滤波跟踪方法和基于积分颜色直方图的观测似然模型.采用积分方向直方图建立一个检测响应图,用其上的观测信息构建建议分布函数,在状态空间中似然概率较大的子空间中进行粒子采样.仿真结果表明,在采用大量粒子跟踪大目标时,提出的跟踪方法的计算复杂度明显低于直方图直接提取的粒子滤波方法, 而且在光照变化的条件下,比采用传统颜色Mean-Shift的跟踪方法准确.     

随机摄动强跟踪粒子滤波算法

如何解决粒子的退化问题和提高算法对突变状态的跟踪能力,是粒子滤波算法研究和应用中需要考虑的两个主要因素.传统的再采样算法虽然可以解决退化问题,但是容易导致粒子耗尽;扩展粒子滤波算法虽然可在一定程度上解决粒子耗尽问题,但其对突变状态的跟踪能力却不近人意;强跟踪粒子滤波算法可以提高对突变状态的跟踪能力,但却未能较好地改善粒子退化问题.针对上述问题,本文将随机摄动再采样方法引入强跟踪粒子滤波算法,提出了一种随机摄动强跟踪粒子滤波算法.当粒子退化问题严重时,对权值最大的粒子迭加随机摄动,用摄动粒子替换退化粒子以解决粒子退化问题,同时由于摄动粒子的加入增加了粒子集的多样性,可在一定程度上缓解粒子耗尽问题,提高算法对突变状态的跟踪能力.利用标准验证模型和分时恒定系统对所提出的算法进行了仿真验证,仿真结果证明了该算法的可行性和有效性.     

基于蝙蝠算法的粒子滤波法研究

标准粒子滤波容易出现粒子贫化问题,滤波精度不稳定,并且需要大量粒子才能对非线性系统进行准确估计,降低了算法的综合性能.针对该问题,本文提出了一种基于蝙蝠算法的新型粒子滤波算法.该算法用粒子表征蝙蝠个体,模拟蝙蝠群体搜索猎物的过程,粒子群体通过调整频率、响度、脉冲发射率,追随当前最优粒子在解空间中进行搜索,并可以动态控制全局搜索及局部搜索的相互转换,进而提髙粒子整体的质量和分布的合理性;此外,改进算法引入Levy飞行策略,从而避免局部极值的不良吸引.实验表明新型粒子滤波方法提高了粒子多样性和滤波预测精度,同时大大降低了对非线性系统进行状态预测所需的粒子数量.     

融合加权均值滤波与流形重构保持嵌入的高光谱影像分类

高光谱影像中波段数过多易导致"维数灾难",而传统高光谱影像维数约简算法仅利用光谱特征而忽略了空间信息.针对上述问题,提出一种融合加权均值滤波与流形重构保持嵌入的维数约简算法.该方法利用影像中地物分布的空间一致性特点,对所有像素进行加权均值滤波,消除同类光谱差异性较大的像素影响,并在流形重构过程中增大空间近邻点的权重,提取出更为有效的鉴别特征,实现维数约简.在PaviaU和Urban高光谱数据集上的实验结果表明:相比于其它相关方法,该方法能获得更高的分类准确度,在分别随机选取5%和1%的训练样本情况下,其总体分类准确度分别提高到98.76%和80.21%.该方法在发现内在低维流形结构的同时,有效融入了影像中的空间信息,改善了分类性能.     

基于不敏粒子滤波的水中非合作磁性目标实时磁定位方法

针对水中非合作磁性目标的实时定位问题, 提出了一种基于不敏粒子滤波(unscented particle filter, UPF)的实时磁定位方法. 从非合作磁性目标的运动特征出发, 建立了状态空间模型, 利用UPF算法对目标状态进行实时估计. 为了提高系统的可观测性, 在算法迭代过程中对粒子状态进行约束及利用最小二乘法反演磁矩. 仿真与铁磁物体定位实验结果表明, 该方法的定位精度较高, 实时定位效果较好, 可用于近场实时磁定位问题中. 关键词： 磁定位 椭球体模型 状态空间模型 不敏粒子滤波     

基于积分直方图的粒子滤波跟踪算法

在目标模型的建立过程中,将积分直方图引入到粒子滤波跟踪框架中,提出了一种快速的颜色直方图计算方法,极大地提高了粒子滤波跟踪算法的实时性。为了进一步提高算法的鲁棒性,引入了一种基于邻域颜色特征的匹配搜索机制,当跟踪精度下降时,能够对跟踪结果进行优化,减小跟踪误差。实验结果表明了该算法的有效性。     

Quantum mechanics on manifolds embedded in Euclidean space

Quantum particles confined to surfaces in higher-dimensional spaces are acted upon by forces that exist only as a result of the surface geometry and the quantum mechanical nature of the system. The dynamics are particularly rich when confinement is implemented by forces that act normal to the surface. We review this confining potential formalism applied to the confinement of a particle to an arbitrary manifold embedded in a higher-dimensional Euclidean space. We devote special attention to the geometrically induced gauge potential that appears in the effective Hamiltonian for motion on the surface. We emphasize that the gauge potential is only present when the space of states describing the degrees of freedom normal to the surface is degenerate. We also distinguish between the effects of the intrinsic and extrinsic geometry on the effective Hamiltonian and provide simple expressions for the induced-scalar potential. We discuss examples including the case of a three-dimensional manifold embedded in a five-dimensional Euclidean space.     

Quantum Sensing of Curvature

We address the problem of sensing the curvature of a manifold by performing measurements on a particle constrained to the manifold itself. In particular, we consider situations where the dynamics of the particle is quantum mechanical and the manifold is a surface embedded in the three-dimensional Euclidean space. We exploit ideas and tools from quantum estimation theory to quantify the amount of information encoded into a state of the particle, and to seek for optimal probing schemes, able to actually extract this information. Explicit results are found for a free probing particle and in the presence of a magnetic field. We also address precision achievable by position measurement, and show that it provides a nearly optimal detection scheme, at least to estimate the radius of a sphere or a cylinder.

     

Momentum and Hamiltonian operators in generalized coordinates

The self-adjointness of momentum operators in generalized coordinates, questioned by Domingos and Caldeira is shown. The momentum operators of a particle and the kinetic part of its Hamiltonian operator constructed from them are characterized as self-adjoint operators and geometrical objects in coordinate-free form. Local coordinates of ann-dimensional Riemannian manifold are taken as the generalized coordinates of the particle. As an example the curvilinear coordinates of Euclidean space are treated. The coefficients of connection and curvature are given on the manifold for which the assumed momentum operators exist. It is found that if our momentum operators form a complete set of mutually commuting observables, the manifold is locally Euclidean, i.e., there exists a local coordinate system such that we obtain the usual Schrödinger correspondence rule.     

辅助变量粒子滤波的声道共振特性轨迹跟踪

提出了一种基于辅助变量粒子滤波技术的连续语音声道共振特性(VTRs)轨迹跟踪方法。该方法基于描述语音信号特征的状态空间模型,采用粒子滤波技术跟踪VTRs的轨迹。语音模型由具有目标导向特性的动态方程和VTRs至倒谱系数(LPCC)的非线性映射构成。该方法有两个特点:首先,采用粒子滤波技术来处理语音模型的非线性问题;其次,在语音模型的状态方程中嵌入辅助变量用于标示VTRs在频域中的分布信息,并为粒子滤波过程中的粒子抽取提供目标导向。实验结果表明,该方法只需少量粒子即可正确跟踪连续语音的VTRs轨迹,而且可以在跟踪过程中避免虚假峰和合并峰的干扰。      

Warped product Kähler manifolds and Bochner–Kähler metrics

Using as an underlying manifold an alpha-Sasakian manifold, we introduce warped product Kähler manifolds. We prove that if the underlying manifold is an alpha-Sasakian space form, then the corresponding Kähler manifold is of quasi-constant holomorphic sectional curvatures with a special distribution. Conversely, we prove that any Kähler manifold of quasi-constant holomorphic sectional curvatures with a special distribution locally has the structure of a warped product Kähler manifold whose base is an alpha-Sasakian space form. As an application, we describe explicitly all Bochner–Kähler metrics of quasi-constant holomorphic sectional curvatures. We find four families of complete metrics of this type. As a consequence, we obtain Bochner–Kähler metrics generated by a potential function of distance in complex Euclidean space and of time-like distance in the flat Kähler–Lorentz space.     

Dimensionality Reduction of SPD Data Based on Riemannian Manifold Tangent Spaces and Isometry

Symmetric positive definite (SPD) data have become a hot topic in machine learning. Instead of a linear Euclidean space, SPD data generally lie on a nonlinear Riemannian manifold. To get over the problems caused by the high data dimensionality, dimensionality reduction (DR) is a key subject for SPD data, where bilinear transformation plays a vital role. Because linear operations are not supported in nonlinear spaces such as Riemannian manifolds, directly performing Euclidean DR methods on SPD matrices is inadequate and difficult in complex models and optimization. An SPD data DR method based on Riemannian manifold tangent spaces and global isometry (RMTSISOM-SPDDR) is proposed in this research. The main contributions are listed: (1) Any Riemannian manifold tangent space is a Hilbert space isomorphic to a Euclidean space. Particularly for SPD manifolds, tangent spaces consist of symmetric matrices, which can greatly preserve the form and attributes of original SPD data. For this reason, RMTSISOM-SPDDR transfers the bilinear transformation from manifolds to tangent spaces. (2) By log transformation, original SPD data are mapped to the tangent space at the identity matrix under the affine invariant Riemannian metric (AIRM). In this way, the geodesic distance between original data and the identity matrix is equal to the Euclidean distance between corresponding tangent vector and the origin. (3) The bilinear transformation is further determined by the isometric criterion guaranteeing the geodesic distance on high-dimensional SPD manifold as close as possible to the Euclidean distance in the tangent space of low-dimensional SPD manifold. Then, we use it for the DR of original SPD data. Experiments on five commonly used datasets show that RMTSISOM-SPDDR is superior to five advanced SPD data DR algorithms.     

区间衍生粒子滤波器

针对非线性、非高斯环境下具有不确定动态模型参数的系统状态估计问题,提出了一种新颖的区间衍生粒子滤波算法.该算法利用区间滤波生成的重要性密度函数,在系统状态转移概率密度的基础上,融入最新的系统观测数据,从而提高了对系统状态后验概率的逼近程度.为了进一步提高算法的实时性,提出一种类似光子衍射的粒子衍生过程,进而缓解了滤波精度与运算量之间的矛盾.通过陀螺/星敏感器组合定姿问题验证了该算法的有效性和鲁棒性.     

混沌信号在无线传感器网络中的盲分离

混沌信号在本质上属于非线性非高斯信号,它在无线传感器网络下的应用还涉及到信号量化问题,这使得混沌信号在此应用环境下的信号盲分离更为棘手.针对此问题,本文在容积卡尔曼粒子滤波的框架下提出一种解决方法.文中首先推导出观测信号的概率密度函数,在量化比特有限的情况下,采用最优量化器,获得最优的量化结果.在此基础上,使用容积卡尔曼滤波器产生粒子滤波中的重要性概率密度函数,融入最新的观测值,提高粒子对系统状态后验概率的逼近,提高信号盲分离的精度.仿真结果表明算法能够有效地分离混合混沌信号,参数估计的精度及其运算量均优于已有的无先导卡尔曼粒子滤波算法,其运行时间为无先导卡尔曼粒子滤波算法的88.77%.