激光雷达地面目标姿态估计

激光雷达距离像能表征目标的几何结构,而同一目标不同姿态的距离图像差别较大,因此目标姿态的准确估计对目标识别尤为重要。Ulf Grenander的模式理论利用变形模板表示观测图像,目标姿态变化引起刚性模板的变形用特殊正交群矩阵表示,在贝叶斯估计框架下,用跳扩散过程从后验密度中取样,以希尔伯特-施密特范数作为误差特性,对目标姿态进行估计,并分析误差性能。针对激光雷达距离像基于地面目标进行研究,利用Grenander的模式理论,在不同载噪比情况下对车辆姿态进行估计。通过对姿态估计误差的分析,表明该算法能够较准确地估计目标的姿态。     

Extended gravitational pose estimation

The model-to-image registration problem is a problem of determining the position and orientation (the pose) of a three-dimensional object with respect to a camera coordinate system. When there is no additional information available to constrain the pose of the object and to constrain the correspondence of object features to image features, the problem is also known as simultaneous pose and correspondence problem, or correspondenceless pose estimation problem. In this paper, we present a new algorithm, called extended gravitational pose estimation (EGPE), for determining the pose and correspondence simultaneously. The algorithm is based on gravitational pose estimation (GPE) algorithm. In our algorithm, the original GPE has been revised to deal with the problem with false image points. For problems with both occluded object points and false image points, we firstly applied single-link agglomerative clustering algorithm to pick out occluded object points when a local minimum has been found, then the revised GPE is applied again on the clustering result to update rotation and translation of the object model. EGPE has been verified on both synthetic images and real images. Empirical results show that EGPE is faster, more stable and reliable than most current algorithms, and can be used in real applications.     

基于自适应像素级注意力模型的场景深度估计

深度估计是传统的计算机视觉任务,在理解三维场景中起着至关重要的作用。基于单目图像的深度估计任务的困难在于如何提取图像特征中大范围依赖的上下文信息,提出了自适应的上下文聚合网络（adaptive context aggregation network,ACANet）用于解决该问题。该方法基于有监督的自注意力模型(supervised self-attention,SSA),能够自适应地学习任意像素之间的具有任务特性的相似性以模拟连续的上下文信息,并通过模型学习的注意力权重分布用来聚合提取的图像特征。将单目深度估计任务设计为像素级的多分类问题,经过设计的注意力损失函数减少RGB图像和深度图的语义不一致性,通过生成的像素级注意力权重对由位置索引的特征进行全局池化。最后提出一种软性有序推理算法（soft ordinal inference,SOI）,充分利用网络的预测置信度,将离散的深度标签转化为平滑连续的深度图,并且提高了准确率（rmse下降了3%）。在公开的单目深度估计基准数据集NYU Depth V2上的实验结果表明:rmse指标为0.490,阈值指标为82.8%,取得了较好的结果,证明了本文提出的算法的优越性。     

基于神经网络的粗糙表面双向反射分布函数建模

分析粗糙表面双向反射分布函数的测量方法,提出一种使用人工神经网络技术建立目标表面材料双向反射分布函数模型的方法。给出测量样品多个入射角度下的BRDF随散射角变化的曲线,从中选取部分曲线输入到神经网络,使用贝叶斯正则化方法训练网络,最终获取双向反射分布函数和入射角、散射角的映射关系模型。使用网络模型计算参与训练和未参与训练的输入角度的散射分布曲线,与实验测量曲线进行比较,结果表明：建立的模型正确,具有应用价值。     

基于点线对应的多摄像机全局位姿估计迭代算法

为了解决工业视觉场景中点和直线段共存情况下的多摄像机位姿估计问题,提出了基于点线特征对应的多摄像机全局位姿估计算法.以正交迭代算法为基础,使用多个固定在平台上的已标定摄像机间的几何约束,以全局方式处理多摄像机的重叠和非重叠视图的点和直线段特征,获得更加稳定精确的位姿估计.该算法先把所有摄像机数据进行统一表达,再把摄像机观测到的全部点特征的目标空间共线性误差和直线段特征的目标空间共面性误差之和作为误差函数,最后经数学推导得到使该误差函数最小化的迭代求解过程.实验结果验证了该算法的有效性以及优越性.     

Pose estimation for UAV aerial refueling with serious turbulences based on extended Kalman filter

In recent years, many pose estimation algorithms were developed, and have been successfully applied to solve unmanned aerial vehicle (UAV) aerial refueling pose estimation problems. This paper mainly focuses on solving this problem under serious turbulences circumstance. The extended Kalman filter is a set of mathematical equations to estimate the state of a process, which is able to support estimations of past, present, and even future states. In reference to previous papers and some simulations, we build up the noise models of refueling boom and atmospheric turbulence. Then, an extend Kalman filter is adopted to solve the pose estimation problem in UAV aerial refueling with serious turbulences. The experimental results demonstrate the feasibility and effectiveness of our proposed approach.     

Comparison of absolute magnitude estimation and relative magnitude estimation for judging the subjective intensity of noise and vibration

The method of magnitude estimation is used in psychophysical studies to obtain numerical values for the intensity of perception of environmental stresses (e.g., noise and vibration). The exponent in a power function relating the subjective magnitude of a stimulus (e.g., the degree of discomfort) to the physical magnitude of the stimulus shows the rate of growth of sensations with increasing stimulus magnitude. When judging noise and vibration, there is no basis for deciding whether magnitude estimation should be performed with a reference stimulus (i.e., relative magnitude estimation, RME) or without a reference stimulus (i.e., absolute magnitude estimation, AME). Twenty subjects rated the discomfort caused by thirteen magnitudes of whole-body vertical vibration and 13 levels of noise, by both RME and AME on three occasions. There were high correlations between magnitude estimates of discomfort and the magnitudes of vibration and noise. Both RME and AME provided rates of growth of discomfort with high consistency over the three repetitions. When judging noise, RME was more consistent than AME, with less inter-subject variability in the exponent, n_s. When judging vibration, RME was also more consistent than AME, but with greater inter-subject variability in the exponent, n_v. When judging vibration, AME may be beneficial because sensations caused by the RME reference stimulus may differ (e.g., occur in a different part of the body) from the sensations caused by the stimuli being judged.     

靶场目标三维姿态测量结果自动修正方法

中轴线法是靶场试验中常用的目标三维姿态测量方法,但该方法求出的俯仰角和偏航角所描述的目标姿态可能与目标实际情况相异,必须进行人工修正。提出一种自动对姿态判读结果进行修正的方法,将目标共线方程与中轴线方程联立求解,可快速地求出角度修正量。实验证明该方法可对中轴线法求解结果进行自动修正,得到正确的目标三维姿态角,提高了姿态判读的速度和可靠性。     

基于影像-位置序图的光学测姿方法

针对空中飞行目标姿态测量问题,提出一种新的光学测姿实现方法,对光学多站测姿实施复杂、单站测姿成像要求高等问题进行改进。该方法充分利用了目标飞行连续影像信息和相应的外弹道位置数据,首先通过目标位置数据确定目标轴线向量起点,根据轴线向量终点的约束方程确定向量终点;然后通过偏航角、俯仰角与轴线向量起点、终点的关系,解算目标姿态角;最后通过姿态连续变化的先验信息,确定解算结果的唯一值。仿真计算和精度分析表明,可实现测姿误差总体小于1,表明提出的方法兼顾了便捷性和测量精度,具有较高工程应用价值。     

孔径扩展中相位修正因子估计的改进算法

传统的扩展拖曳阵列尺寸算法(Extended Towed Array Measurement,ETAM)在信噪比不够高、相位修正因子的相位角处于以间断点为中心的"跳变区间"时,扩展孔径均有可能失效。针对传统的ETAM算法中互相关相位角的间断点导致的算法不稳定问题,提出了一种相位修正因子估计的改进算法。该算法使用归一化互相关复向量的统计平均值作为相位修正因子的最小二乘估计,消除了互相关相位角的间断点带来的不利影响。数值仿真和实验数据分析结果表明:对于相位稳定甚至相位随机的单目标信号,改进算法均能有效扩展孔径,获得相对于常规波束形成(Conventional Beamforming,CBF)更高的方位分辨率和检测信噪比;相比于传统的ETAM算法,改进算法提高了相位修正因子的估计准确性,从而有效提高了算法的稳定性。      

基于人脸姿态估计的虚拟眼镜试戴技术

为解决用户线上眼镜的最优选购,提出了一种基于人脸姿态估计的虚拟眼镜试戴技术。首先采用肤色模型与形状模型结合的算法对场景中的人脸区域进行检测,然后根据人眼在人脸中的几何位置关系实现人眼的精确定位;进一步利用人眼对称性先验知识来估计脸在三维空间中的姿态信息,即人脸与正面的偏移角度;最后,依据人眼位置和人脸姿态将眼镜图像融合到眼睛区域,即完成眼镜的虚拟试戴。该方法为3D环境下客户与商品之间虚拟视觉化的实现提供了一种可靠的技术支撑和应用思路。     

On a recursive method for the estimation of unknown parameters of partially observed chaotic systems

We investigate a recently proposed method for on-line parameter estimation and synchronization in chaotic systems. This novel technique has been shown effective to estimate a single unknown parameter of a primary chaotic system with known functional form that is only partially observed through a scalar time series. It works by periodically updating the parameter of interest in a secondary system, with the same functional form as the primary one but no explicit coupling between their dynamic variables, in order to minimize a suitably defined cost function. In this paper, we review the basics of the method, and investigate its robustness and new extensions. In particular, we study the performance of the novel technique in the presence of noise (either observational, i.e., an additive contamination of the observed time series, or dynamical, i.e., a random perturbation of the system dynamics) and when there is a mismatch between the primary and secondary systems. Numerical results, including comparisons with other techniques, are presented. Finally, we investigate the extension of the original method to perform the estimation of two unknown parameters and illustrate its effectiveness by means of computer simulations.     

基于点和直线段对应的扩展正交迭代位姿估计算法

正交迭代算法是基于点特征,全局且快速收敛的位姿估计算法,是目前性能最优的实时位姿估计算法之一.对摄像机观测到的直线段提出了其不确定性描述的新方法:目标空间直线段误差,并把该误差融入到正交迭代算法的位姿计算过程中,形成扩展正交迭代算法.该算法可以同时利用点和直线段特征.算法先把点共面性方程表达为与正交迭代算法中点共线性方程一致的数学形式,再根据目标空间点共面性误差定义了目标空间直线段共面性误差.接着把目标空间点共线性误差和直线段共面性误差二者之和作为误差函数.最后推导出使该函数最小化的迭代求解过程.实验结果表明,本文算法是有效的,精确的.与正交迭代算法的计算结果对比,由于可以同时利用直线段特征,扩展正交迭代算法的位姿估计误差更低,抗噪声性能更强,算法更稳定.     

Motion blur parameters estimation for image restoration

This paper deals with estimation of parameters for motion blurred images. The objectives are to estimate the length (L) and the blur angle (θ) of the given degraded image as accurately as possible so that the restoration performance can be optimised. Gabor filter is utilized to estimate the blur angle whereas a trained radial basis function neural network (RBFNN) estimates the blur length. Once these parameters are estimated the conventional restoration is performed. To validate the proposed scheme, simulation has been carried out on standard images as well as in real images subjected to different blur angles and lengths. The robustness of the scheme is also validated in noise situations of different strengths. In all situations, the results have been compared with standard schemes. It is in general observed that the proposed scheme outperforms its counterparts in terms of restoration parameters and visual quality.     

Power-Modified Kies-Exponential Distribution: Properties,Classical and Bayesian Inference with an Application to Engineering Data

We introduce here a new distribution called the power-modified Kies-exponential (PMKE) distribution and derive some of its mathematical properties. Its hazard function can be bathtub-shaped, increasing, or decreasing. Its parameters are estimated by seven classical methods. Further, Bayesian estimation, under square error, general entropy, and Linex loss functions are adopted to estimate the parameters. Simulation results are provided to investigate the behavior of these estimators. The estimation methods are sorted, based on partial and overall ranks, to determine the best estimation approach for the model parameters. The proposed distribution can be used to model a real-life turbocharger dataset, as compared with 24 extensions of the exponential distribution.     

A New Generator of Probability Models: The Exponentiated Sine-G Family for Lifetime Studies

In this article, we propose the exponentiated sine-generated family of distributions. Some important properties are demonstrated, such as the series representation of the probability density function, quantile function, moments, stress-strength reliability, and Rényi entropy. A particular member, called the exponentiated sine Weibull distribution, is highlighted; we analyze its skewness and kurtosis, moments, quantile function, residual mean and reversed mean residual life functions, order statistics, and extreme value distributions. Maximum likelihood estimation and Bayes estimation under the square error loss function are considered. Simulation studies are used to assess the techniques, and their performance gives satisfactory results as discussed by the mean square error, confidence intervals, and coverage probabilities of the estimates. The stress-strength reliability parameter of the exponentiated sine Weibull model is derived and estimated by the maximum likelihood estimation method. Also, nonparametric bootstrap techniques are used to approximate the confidence interval of the reliability parameter. A simulation is conducted to examine the mean square error, standard deviations, confidence intervals, and coverage probabilities of the reliability parameter. Finally, three real applications of the exponentiated sine Weibull model are provided. One of them considers stress-strength data.     

基于矢量化差分相位的单分布源解耦二维波达方向估计

在分布源(包括相干分布源和非相干分布源)的二维波达方向估计中,均匀圆阵由于可实现全方位测角、具有较高的分辨率,得到了广泛的应用,然而现有的估计算法均需要谱峰搜索和特征值分解,复杂度较高.针对此问题,考虑单个相干分布源或非相干分布源入射两种情况,提出了一种基于矢量化差分相位的解耦二维波达方向快速估计算法.该算法首先基于空间频率近似模型,证明了任意单个分布源入射时,均匀圆阵中不同阵元接收信号间的差分相位均不受角度扩展参数的影响;基于此特性,通过获取差分相位即可实现中心波达角的解耦合;接下来,提取采样协方差矩阵的严格上三角元素相位,即对应于各阵元间的差分相位,并进行矢量化处理,最终将波达方向估计问题转化为一个最小二乘问题,从而直接得到闭式解,避免了谱峰搜索和特征值分解运算,大幅度降低了复杂度.理论分析和仿真实验表明,所提算法具有较高的估计精度,并且无需角信号分布的先验信息,同时具备较低的计算复杂度和硬件复杂度,有利于复杂环境下阵列测向等工程实践.     

缩短两级爆磁压缩发生器输出脉冲宽度的一种途径

 缩短两级爆磁压缩发生器输出脉冲宽度的一种有效办法是增大第二级螺旋线圈的锥角,但增大锥角会降低发生器电感压缩比,合理的锥角设计是此类发生器的关键。介绍了三种能带大电感负载、结构紧凑的两级爆磁压缩发生器,第一级结构参数相同,第二级螺旋线圈锥角分别为6°,8°,10°。在电感负载上进行的实验结果表明：其输出电流分别达到140,90和50 kA,电流脉冲宽度分别为12.5,8.5和6.3 μs,对应的名义输出功率分别为7,7和3 GW。     

基于红外和可见光图像逐级自适应融合的场景深度估计

从图像中恢复场景的深度是计算机视觉领域中的一个关键问题。考虑到单一类型图像在深度估计中受场景不同光照的限制,提出了基于红外和可见光图像逐级自适应融合的场景深度估计方法（PF-CNN）。该方法包括双流滤波器部分耦合网络、自适应多模态特征融合网络以及自适应逐级特征融合网络。在双流卷积中红外和可见光图像的滤波器部分耦合使两者特征得到增强;自适应多模态特征融合网络学习红外和可见光图像的残差特征并将两者自适应加权融合,充分利用两者的互补信息;逐级特征融合网络学习多层融合特征的结合,充分利用不同卷积层的不同特征。实验结果表明:PF-CNN在测试集上获得了较好的效果,将阈值指标提高了5%,明显优于其他方法。