Two-dimensional regularized inversion of AMT data based on rotation invariant of Central impedance tensor

Considering the uncertainty of the electrical axis for two-dimensional audo-magnetotelluric(AMT) data processing, an AMT inversion method with the Central impedance tensor was presented. First, we present a calculation expression of the Central impedance tensor in AMT, which can be considered as the arithmetic mean of TE-polarization mode and TM-polarization mode in the twodimensional geo-electrical model. Second, a least-squares iterative inversion algorithm is established, based on a smoothnessconstrained model, and an improved L-curve method is adopted to determine the best regularization parameters. We then test the above inversion method with synthetic data and field data. The test results show that this two-dimensional AMT inversion scheme for the responses of Central impedance is effective and can reconstruct reasonable two-dimensional subsurface resistivity structures. We conclude that the Central impedance tensor is a useful tool for two-dimensional inversion of AMT data.

A Secure Rotation Invariant LBP Feature Computation in Cloud Environment

In the era of big data,outsourcing massive data to a remote cloud server is a promising approach.Outsourcing storage and computation services can reduce storage costs and computational burdens.However,public cloud storage brings about new privacy and security concerns since the cloud servers can be shared by multiple users.Privacy-preserving feature extraction techniques are an effective solution to this issue.Because the Rotation Invariant Local Binary Pattern(RILBP)has been widely used in various image processing fields,we propose a new privacy-preserving outsourcing computation of RILBP over encrypted images in this paper(called PPRILBP).To protect image content,original images are encrypted using block scrambling,pixel circular shift,and pixel diffusion when uploaded to the cloud server.It is proved that RILBP features remain unchanged before and after encryption.Moreover,the server can directly extract RILBP features from encrypted images.Analyses and experiments confirm that the proposed scheme is secure and effective,and outperforms previous secure LBP feature computing methods.

Form Invariance and Noether Symmetries of Rotational Relativistic Birkhoff Systems

Under the infinitesimal transformations of groups, a form invariance of rotational relativistic Birkhoffsystems is studied and the definition and criteria are given. In view of the invariance of rotational relativistic PfaffBirkhoff D'Alcmbert principle under the infinitesimal transformations of groups, the theory of Noether symmetries ofrotational relativistic Birkhoff systems are constructed. The relation between the form invariance and the Noethersymmetries is studied, and the conserved quantities of rotational relativistic Birkhoff systems are obtained.

Form Invariance and Noether Symmetries of Rotational Relativistic Birkhoff Systems

Under the infinitesimal transformations of groups, a form invariance of rotational relativistic Birkhoff systems is studied and the definition and criteria are given. In view of the invariance of rotational relativistic Pfaff Birkhoff D'Alembert principle under the infinitesimal transformations of groups, the theory of Noether symmetries of rotational relativistic Birkhoff systems are constructed. The relation between the form invariance and the Noether symmetries is studied, and the conserved quantities of rotational relativistic Birkhoff systems are obtained.

Content-Based Image Retrieval with Feature Extraction and Rotation Invariance

Over recent years, Convolutional Neural Networks (CNN) has improved performance on practically every image-based task, including Content-Based Image Retrieval (CBIR). Nevertheless, since features of CNN have altered orientation, training a CBIR system to detect and correct the angle is complex. While it is possible to construct rotation-invariant features by hand, retrieval accuracy will be low because hand engineering only creates low-level features, while deep learning methods build high-level and low-level features simultaneously. This paper presents a novel approach that combines a deep learning orientation angle detection model with the CBIR feature extraction model to correct the rotation angle of any image. This offers a unique construction of a rotation-invariant CBIR system that handles the CNN features that are not rotation invariant. This research also proposes a further study on how a rotation-invariant deep CBIR can recover images from the dataset in real-time. The final results of this system show significant improvement as compared to a default CNN feature extraction model without the OAD.

Exact Solutions and Invariant Sets to General Reaction-Diffusion Equation

In this paper, we introduce new invariant sets, and the invariant sets and exact solutions to general reactiondiffusion equation are discussed. It is shown that there exist a class of exact solutions to the equations that belong to the invariant sets.

Ordered Rate Constitutive Theories for Non-Classical Thermoviscoelastic Fluids with Internal Rotation Rates

The paper presents constitutive theories for non-classical thermoviscoelastic fluids with dissipation and memory using a thermodynamic framework based on entirety of velocity gradient tensor. Thus, the conservation and the balance laws used in this work incorporate symmetric as well as antisymmetric part of the velocity gradient tensor. The constitutive theories derived here hold in coand contra-variant bases as well as in Jaumann rates and are derived using convected time derivatives of Green’s and Almansi strain tensors as well as the Cauchy stress tensor and its convected time derivatives in appropriate bases. The constitutive theories are presented in the absence as well as in the presence of the balance of moment of moments as balance law. It is shown that the dissipation mechanism and the fading memory in such fluids are due to stress rates as well as moment rates and their conjugates. The material coefficients are derived for the general forms of the constitutive theories based on integrity. Simplified linear (or quasi-linear) forms of the constitutive theories are also presented. Maxwell, Oldroyd-B and Giesekus constitutive models for non-classical thermoviscoelastic fluids are derived and are compared with those derived based on classical continuum mechanics. Both, compressible and incompressible thermoviscoelastic fluids are considered.

Invariant Sets and Exact Solutions to Nonlinear Diffusion Equations with x-Dependent Convection and Absorption

In this paper, we introduce a new invariant set Eo={u：ux=f＇（x）F（u）＋ε[g＇（x）-f＇（x）g（x）]F（u）×exp（-∫^u1/F（z）dz）}where f and g are some smooth functions of x, ε is a constant, and F is a smooth function to be determined. The invariant sets and exact sohltions to nonlinear diffusion equation ut = （ D（u）ux）x ＋ Q（x, u）ux ＋ P（x, u）, are discussed. It is shown that there exist several classes of solutions to the equation that belong to the invariant set Eo.

基于总体最小二乘-旋转不变算法的地表核磁共振信号参数估计

在地表核磁共振(SNMR)找水系统中,根据SNMR信号的参数能够预估地下含水层的储水量、导电性以及孔隙结构等信息。然而在实际应用中探测现场采集的SNMR信号十分微弱,易受到环境噪声干扰,导致无法直接获取SNMR信号的参数。针对这一问题,该文提出基于总体最小二乘-旋转不变法(TLS-ESPRIT)的地表核磁共振信号参数估计方法。基于谐波噪声与SNMR信号的相似信号特征构成一个由多个正弦衰减信号叠加的混合信号模型,使用TLS-ESPRIT将混合信号参数提取问题转换为旋转不变矩阵的广义特征值求解,从而获得SNMR信号的拉莫尔频率和弛豫时间,并结合最小二乘法求得其初始振幅和相位。仿真信号和实测信号实验结果表明此方法能够估计出混有随机噪声和工频谐波噪声的SNMR信号的参数,相比传统的谐波建模方法,在参数提取精度上效果更好。

基于区域渐进校准网络的人脸检测与定位

为解决角度变化下的人脸检测中存在参数量大及角度幅度变量小的问题,提出区域渐进校准网络用于任意平面角度的人脸检测,通过级联网络结构降低角度变化、提升网络运行速度。采用区域生成网络产生高质量的候选区域,构造渐进校准网络,逐步缩小面部平面角度变化范围,同时由粗到细地对候选区域执行面部检测。其中,特征提取的中间层融合参数量较少时,更好地表示了面部特征,调整锚的设置解决小尺度面部问题。在角度增强的FDDB(face detection data set and benchmark)数据集与WIDER FACE数据集上的实验结果表明,提出的方法分别取得了89.1%与90.4%的平均召回率,准确度高于快速区域卷积神经网络(Faster RCNN),且运行速度更快。在实际项目中使用该算法,验证了该方法的有效性及可行性。

二维磁流体方程的高分辨率旋转通量格式

若待解方程满足旋转不变性,则可通过旋转通量法有效消除近似Riemann求解器的激波不稳定现象,抑制非物理现象的产生。针对二维理想磁流体(MHD)方程和浅水波磁流体(SWMHD)方程,构造了通量函数的类旋转矩阵,给出了方程的旋转不变性证明;根据该性质对控制方程做类一维处理,推导了方程的半离散旋转通量格式;利用通量限制器,将熵稳定通量和反扩散通量进行加权组合,得到能够自适应调整耗散量的高分辨率旋转通量格式。数值实验表明,此格式能精确捕捉解的结构,分辨率高、鲁棒性强,且易向高维推广。

具有变/定转轴的一类分岔2Rv广义并联机构构型综合

现有大部分2R并联机构靠近定平台的第1个转轴方向矢量不变,靠近动平台的第2个转轴方向矢量只随着第1个转轴而变化,第2转轴相对于动平台不变。本文利用有限旋量理论,在具有变/定转轴分岔1Rv(Rv表示变转轴转动)并联机构基础上,提出一类具有变/定轴线的2Rv并联机构。分析了机构装配条件和驱动配置。此种分岔2Rv并联机构包含4种运动模式,即定-定转轴运动模式、定-变转轴运动模式、变-定转轴运动模式和变-变转轴运动模式。将传统的2条定转轴2R并联机构拓展为具有变/定转轴(变转轴和定转轴)的分岔2Rv广义并联机构。

考虑夯锤姿态特性的夯沉量单目视觉测量方法研究

夯沉量是反映强夯地基加固质量的关键指标,人工测量方法不仅投入高、效率低下,且施工安全无法保障。为此,基于单目摄影测量提出一种强夯夯沉量非接触测量方法,实现了强夯夯沉量的高效测算。深入分析并科学定义了适用于单目视觉监测的夯沉量定义;基于夯锤标识的旋转不变性,提取表征夯锤姿态的特征点几何信息,结合视觉成像基本原理,构建了测算夯锤三维姿态的数值方程,求解得到夯锤特征点的三维坐标;在施工现场对所提方法进行工程实测,并通过多余观测对解算结果进行优化尝试。试验及工程应用结果表明,本方法具有精度高、速度快、稳健性强等特点,可有效应用于强夯施工质量监测关键指标参数夯沉量的自动解算,为当前强夯夯沉量自动化监测技术的进步提供了一个新的方向和手段。

基于深度旋转不变特征图哈希的遥感图像检索

哈希技术采用紧致哈希码表示数据,因其高效性被广泛应用于大规模遥感图像检索任务。受卫星观测影响,同一地物在不同遥感图像中呈现不同角度,导致检索性能下降。为解决该问题,该文提出深度特征图旋转不变哈希方法(DRIFMH),包括特征提取、哈希量化2个模块。特征提取模块对特征图进行不同角度旋转,提出特征一致性损失,使不同旋转角度的图像特征保持一致,克服旋转带来的不利影响。哈希量化模块对图像特征进行二值量化,生成哈希码,引入分类交叉熵损失,提升哈希码的鉴别能力。该文选取经典遥感图像数据集AID、UCMD作为实验数据集,将DRIFMH与多个哈希方法进行实验对比,结果表明DRIFMH能够生成旋转不变的遥感图像特征,提升大规模遥感图像检索性能。

SR-Det:面向工业场景下细长和旋转目标的鲁棒检测

目标检测广泛应用于工业领域,譬如缺陷检测。然而,在检测过程中依然存在任意旋转和大宽高比问题。一是水平锚框方法难以准确地定位物体;二是卷积神经网络(Convolutional Neural Networks,CNNs)在提取特征时表现不佳;三是普通的损失函数对细长的目标不敏感。针对上述问题,本文研究了SR-Det (Slender and Rotated Detecto)模型,包含以下3个部分。首先是旋转区域校准(Rotated Region Calibration,RRC)模块。该算法以不同大小和宽高比的水平提议作为输入,以相应的旋转提议作为输出。然后是旋转角度提议对齐模块(Rotated Angle Proposal Align,RAP-Align)来保证特征信息的质量。最后是基于交并比(Intersection Over Union,IoU)策略的R-IoU函数(Rotated Intersection Over Union)以指导模型最大化预测框和GT (Ground Truth)框之间的重叠面积。实验证明,本文提出的方法在金属罐数据集和幕墙数据集上取得了最优的效果,证明了该方法的有效性。

基于融合式PC-ORB的异源图像配准算法

异源图像配准中,由于图像的成像机理差异,图像像素强度关联和旋转畸变是不可避免的两大问题,针对图像像素强度关联问题,提出了基于辐射不变特征变换(radiation-variation insensitive feature transform,RIFT)的图像配准算法,对图像间像素关联差异小的图像对配准有良好的精度,但对旋转畸变图像会产生较多错误匹配。对于旋转畸变问题,传统的ORB(oriented fast and rotated brief)算法,对旋转图像的配准有一定的稳定性,但对于强度变化不明显的图像对,特征点检测质量较低,配准精度不理想。因此本文将相位一致性(phase consistency,PC)融合进ORB算法,利用相位信息代替传统的图像强度信息,再构造旋转不变性BRIEF特征描述子,对图像像素强度变化和旋转畸变均具有鲁棒性。用图像像素强度关联不明显的红外图像和可见光图像进行配准实验,本文算法针对不同旋转幅度的图像的配准精度较高,RMSE稳定在1.7~2.1,优于RIFT算法,在特征点检测数量、配准精度和效率等性能上均有良好性能。

融合旋转不变约束的哈希分类器的设计与应用

分类器是目标识别、目标检测的重要部分,而针对遥感图像的目标检测在交通、军事、农业等方面具有重要的应用价值。但随着遥感图像的分辨率急剧增加,如何提高遥感图像目标检测的效率成为一个挑战。利用遥感目标具有的旋转不变特性以及哈希学习具有的快速分类能力,设计并实现了融合旋转不变约束的哈希分类器,其目的是使遥感目标在旋转前后具有相似的二进制哈希码。通过实验表明该分类器能在大幅提高检测速度的同时提高检测的准确性,并且可以拓展到其他哈希学习方法上。

基于信息熵的二维局部二值模式静脉识别

基于现有LBP算法及其变体无法提取图像高维特征的问题,提出一种基于信息熵的二维局部二值模式识别算法。此方法首先利用统一局部二值模式(ULBP)对图像进行低维特征的提取,随后将图像信息熵与统一局部二值模式图谱进行结合获取熵值加权的统一局部二值模式图谱(EULBP),并利用滑动窗口实现对局部区域内模式间共现特征信息的统计,以其结果作为图像特征表达。并以直方图交叉距离为基础构建模式分类器,验证其识别性能。实验结果表明,在SDUMLA⁃HMT数据集以及马来西亚理工大学指静脉数据集(FV⁃USM)中,提出的算法能取得99.94%和98.84%的平均识别率。

MULTI-INVARIANCE ESPRIT-LIKE ALGORITHMS FOR COHERENT DOA ESTIMATION

Estimation of Signal Parameters via Rotational Invariance Technique(ESPRIT) algorithm can estimate Direction-Of-Arrival(DOA) of coherent signal,but its performance can not reach full satisfaction.We reconstruct the received signal to form data model with multi-invariance property,and multi-invariance ESPRIT algorithm for coherent DOA estimation is proposed in this paper.The proposed algorithm can resolve the DOAs of coherent signals and performs better in DOA estimation than that of ESPRIT-like algorithm.Meanwhile,it identifies more DOAs than ESPRIT-like algorithm.The simulation results demonstrate its validity.

Rotation Scaling and Translation Invariants of 3D Radial Shifted Legendre Moments

This paper proposes a new set of 3D rotation scaling and translation invariants of 3D radially shifted Legendre moments. We aim to develop two kinds of transformed shifted Legendre moments： a 3D substituted radial shifted Legendre moments （3DSRSLMs） and a 3D weighted radial one （3DWRSLMs）. Both are centered on two types of polynomials. In the first case, a new 3D ra- dial complex moment is proposed. In the second case, new 3D substituted/weighted radial shifted Legendremoments （3DSRSLMs/3DWRSLMs） are introduced using a spherical representation of volumetric image. 3D invariants as derived from the sug- gested 3D radial shifted Legendre moments will appear in the third case. To confirm the proposed approach, we have resolved three is- sues. To confirm the proposed approach, we have resolved three issues： rotation, scaling and translation invariants. The result of experi- ments shows that the 3DSRSLMs and 3DWRSLMs have done better than the 3D radial complex moments with and without noise. Sim- ultaneously, the reconstruction converges rapidly to the original image using 3D radial 3DSRSLMs and 3DWRSLMs, and the test of 3D images are clearly recognized from a set of images that are available in Princeton shape benchmark （PSB） database for 3D image.