求解分数傅里叶变换衍射积分的一种快速算法

在对Lohmann 二型分数傅里叶变换(FRT)和菲涅耳衍射积分进行比较的基础上,给出基于快速傅里叶变换(FFT)求解该分数傅里叶变换和菲涅耳衍射积分的快速算法及算法适用范围.数值模拟实验证明了理论的可靠性和算法的高效性.此快速算法为分数傅里叶变换在工程实际中的进一步广泛应用奠定了基础.     

基于快速傅里叶变换的实时相移校正算法

在相移测量剖面术（ＰＭＰ）中,提出了一种基于快速傅里叶变换的相移校正算法,利用最初多采的一幅条纹图的信息和傅里叶变换调制度模板,快速地计算出当前的相移值。采用该算法,能实时准确地控制投影光栅的移动,大大提高了测量系统的工和自动化程度,具有极大的实用价值。     

基于灰度统计的快速模板匹配算法

针对传统的基于灰度值的匹配算法计算量大、时间复杂度高的缺陷,提出了一种基于灰度统计的快速模板匹配算法,以十字特征直线代替模板进行粗匹配,仅对搜索图中像素点数目较少的一段灰度区域进行匹配,从而大大减小了计算量,使匹配速度得到极大提高.仿真实验证明本算法匹配速度快、较传统的灰度相关算法快了二至三个数量级,具有较强鲁棒性.     

干涉条纹快速加窗傅里叶滤波方法的研究

滤波是干涉图像处理中的关键问题之一。为解决加窗傅里叶滤波在干涉条纹处理中运行时间长的问题,采用了快速傅里叶变换算法用以减少计算时间。通过加窗傅里叶变换,克服了傅里叶变换在干涉图相位局部信息不能提取的缺点,提出了一种快速加窗傅里叶滤波方法,实现了干涉条纹的快速滤波。在保证滤波效果的同时,缩短了计算时间,大大提高了干涉条纹处理速度,从根本上解决了加窗傅里叶滤波耗时长的问题,使得加窗傅里叶滤波方法应用于散斑在线检测成为了可能。实验验证了本方法的快速性、有效性和可靠性。     

一种改进的红外图像归一化互相关匹配算法

分析了传统归一化互相关算法在红外空中目标匹配定位时失效的原因,提出一种改进的红外图像归一化互相关匹配算法.该方法将模板和匹配区域之间的纹理相关计算看作一个最优化问题,寻求使图像纹理相关匹配鲁棒性最好的相关基准值,用图像的相关基准函数替代传统方法中的区域均值部分,构造了一种适用于的红外目标匹配的归一化相关算法.实验结果表明,该相关匹配算法对模板中背景部分的变化和非均匀性亮度变化有良好的抗干扰能力,较好地解决了恶劣环境下红外对空目标跟踪中匹配定位出错的问题.     

一种改良的啁啾变换算法及其应用

快速啁啾算法引入两次快速傅里叶变换（FFT）及一个解析高斯核,计算复杂度低于卷积算法.通过对啁啾算法实现过程进行的改进,避免了该算法在实现过程中存在的一些问题,比如输出窗口小、信号丢失、计算复杂度稍大等缺点. 把算法用于简单的可求得解析解的系统并与之做比较. 对高斯函数,最大误差通常在10^-15数量级,而对矩形函数,由于受FFT算法计算精度的影响,误差在10^-3数量级,但这并不影响算法的性能. 最后把算法用于一种典型的标量衍射系统及分数傅里叶变换的计算,获得了很好的结果. 关键词： 快速啁啾算法 啁啾Z变换 菲涅耳变换 分数傅里叶变换     

基于环向特征的快速旋转不变性匹配算法

景象匹配中经常会出现基准图与模板图之间存在任意角度旋转的情况,传统的结构化模板匹配算法是通过逐个比较基准图与模板图对应像素灰度值之间的相关性来进行匹配的,所以只能工作在小角度旋转的情况下,对任意角度旋转的景象匹配是无能为力的.利用圆的各向同性和投影特征提出了圆环投影匹配算法,并构造了环向特征投影不变量,用环向特征作匹配.实验结果表明,在同等条件下,该方法较圆投影算法具有更高的鲁棒性,且匹配速度可增快7倍,匹配概率可提高10%～20%.     

大光程差PEM-FTS的快速光谱重建

弹光调制傅里叶变换光谱仪(PEM-FTS)的调制光程差是高速、非线性变化,每秒可产生上万张干涉图。为了实现高速等时间采样干涉信号的快速光谱反演,对大光程差弹光调制干涉信号的特性、加速非均匀快速傅里叶变换算法(NUFFT)进行研究。加速非均匀快速傅里叶变换算法是基于卷积核函数插值的快速傅里叶变换算法,此算法的核函数类型、参数τ、延伸影响因子q、过采样率μ等参数的选择对算法准确度以及复杂度有影响。在分析这些参数对算法影响的基础上,在μ=2,q=10,τ=1×10-6时,将加速的NUFFT算法应用于弹光调制傅里叶变换光谱仪中,重建了632.8 nm的激光和氙灯光谱,复原的632.8 nm激光光谱的频率偏差小于0.013 52,插值时间小于0.267 s。实验表明加速的NUFFT算法有较快的运行速度和小的频率偏差,能快速准确地重建大光程差PEM-FTS的光谱。     

分数傅里叶域数字水印算法

基于离散分数傅里叶变换(DFRFT)快速算法发展了一种分数傅里叶谱域图像水印算法.该算法根据分数傅里叶变换谱具有空域和频域双域信息表达能力,分别对原始图像和所加水印信息进行不同级次的分数傅里叶变换,提取水印分数傅里叶谱的低频成分并直接将其叠加到原始图像的分数傅里叶谱中的对角像元上,然后再进行逆变换得到水印图像.在JPEG压缩、图像旋转及剪切等攻击方式下,对该水印算法进行了鲁棒性分析,数值实验表明该水印算法具有良好的抗攻击性和安全性.     

基于位相相关检测的图像配准算法

分析了传统的傅里叶-梅林变换和最近提出的虚拟极坐标傅里叶变换算法在户外图像注册中的局限性,提出了一种更加适合户外摄像机姿态注册的基于对数傅里叶变换的图像配准算法。该算法能够独立地计算沿两坐标轴方向的不同缩放因子和计算两图像间较大的旋转角度,扩展了基于傅里叶变换的图片配准算法的实际应用范围。在关键技术研究方面,提出了由升余弦函数构成的低通滤波器和一种快速收敛的角度搜索技术,使该算法达到了很高的效率和鲁棒性。     

Thin Films Made Fast and Modified Fast

Hyperfine Interactions - Thin films are playing a more and more important role for technological applications and there are many aspects of materials surface processing and thin film production,...     

Fast fission phenomenon

We review some experimental results which cannot be understood within the usual concepts of dissipative collisions. We present a dynamical model in which fast fission phenomenon appears in a natural way if certain conditions are fulfilled. Two kinds of fast fission are described, according to the size of the system. An extended notion of the so called fusion process is proposed and calculations of fusion cross sections are compared with experimental data. The fluctuations associated with the collective variables which are involved during fast fission and compound nucleus fission are treated within the framework of the linear response theory. The study of mass distributions supports the existence of fast fission with medium systems as an intermediate mechanism between the deep inelastic collisions and the compound nucleus formation.     

Fast proton decay

We consider proton decay in the testable flipped SU(5)×UX₍₁₎$\mathit{SU}(5)\times U{(1)}_X$ models with TeV-scale vector-like particles which can be realized in free fermionic string constructions and F-theory model building. We significantly improve upon the determination of light threshold effects from prior studies, and perform a fresh calculation of the second loop for the process p→e⁺π⁰$p\to e^{+}{\pi }^0$ from the heavy gauge boson exchange. The cumulative result is comparatively fast proton decay, with a majority of the most plausible parameter space within reach of the future Hyper-Kamiokande and DUSEL experiments. Because the TeV-scale vector-like particles can be produced at the LHC, we predict a strong correlation between the most exciting particle physics experiments of the coming decade.     

Fast deflagration-to-detonation transition

Studies of fast deflagration-to-detonation transition in gas and drop air-fuel explosive mixtures are reviewed. Fast deflagration-to-detonation transition is understood as the appearance of detonation at which a turbulent flame is sped up to a much lower velocity than that required for the classic deflagration-to-detonation transition in a straight tube with smooth or rough walls. The main goal of studies was to determine conditions under which fast deflagration-to-detonation transition was possible in weakly sensitive explosive mixtures at very low ignition energies. Examples of fast deflagration-to-detonation transitions checked experimentally and by multidimensional numerical calculations are given, including deflagration-to-detonation transitions (1) in a tube segment with regular obstacles of a special shape, (2) in tube coils, and (3) in tubes with U-shaped bends. In all cases, fast deflagration-to-detonation transition occurs because of the formation of distributed ignition zones in reflections of a running shock wave formed by an accelerated flame. The use of various combinations of reflecting elements can induce fast deflagration-to-detonation transition in an air mixture of aviation kerosene at ignition energies at a level of 5 J.     

快速光造型技术

本文介绍的快速光造型技术以新颖的“迭加”成型法取代传统的“去除”成型法，综合运用了激光、高分子材料、模具ＣＡＤ／ＣＡＭ及自动控制等学科技术。     

Fast topological imaging

Mathematical optimization methods based on the topological sensitivity analysis have been used to develop innovative ultrasonic imaging methods. With a single illumination of the medium, they have proved experimentally to yield a lateral resolution comparable to classical multiple-illumination techniques. As these methods are based on the numerical simulations of two wave fields, they require extensive computation. A time-domain finite-difference scheme is usually used for that purpose. This paper presents the development of an experimental imaging method based on the topological sensitivity. The numerical cost is reduced by replacing the numerical simulations by simple mathematical operations between the radiation patterns of the array’s transducers and the frequency-domain signals to be emitted. These radiation patterns are preliminary computed once and for all. They were obtained with a finite element model for the anisotropic elastodynamic case and with semi-analytical integrations for the acoustic case. Experimental results are presented for a composite material sample and for a prefractal network immersed in water. A lateral resolution below 2.5 times the wavelength is obtained with a single plane wave illumination. The method is also applied with multiple illuminations, so that objects hidden in complex media can be investigated.     

Fast dissipative collisions

At a bombarding energy of 14.7 MeV/u the charge, mass and angular distributions have been measured as a function of energy loss for reaction products induced by ⁹⁸Mo and ⁹²Mo projectiles on targets of ⁹⁸Mo, ¹⁴⁷Sm, ¹⁵⁴Sm and ⁹²Mo, ¹⁵⁴Sm, ²³⁸U, respectively. An analysis of the product distributions shows the presence of a non-equilibrium sharing of the excitation energy, an exponential growth of the charge and mass variances, a complete $\text{N}\text{Z}$ equilibration and a large yield of non-binary complex fragments. Monte Carlo simulations are found to be an indispensable tool of the analysis.