Active contours with normally generalized gradient vector flow external force

Gradient vector flow （GVF） is an effective external force for active contours, but its iso- tropic nature handicaps its performance. The recently proposed gradient vector flow in the normal direction （NGVF） is anisotropic since it only keeps the diffusion along the normal direction of the isophotes; however, it has difficulties forcing a snake into long, thin boundary indentations. In this paper, a novel external force for active contours called normally generalized gradient vector flow （NGGVF） is proposed, which generalizes the NGVF formulation to include two spatially varying weighting functions. Consequently, the proposed NGGVF snake is anisotropic and would improve ac- tive contour convergence into long, thin boundary indentations while maintaining other desirable properties of the NGVF snake, such as enlarged capture range, initialization insensitivity and good convergence at concavities. The advantages on synthetic and real images are demonstrated.

Hybrid gradient vector fields for path-following guidance

Guidance path-planning and following are two core technologies used for controlling un-manned aerial vehicles(UAVs)in both military and civilian applications.However,only a few approaches treat both the technologies simultaneously.In this study,an innovative hybrid gradient vector fields for path-following guidance(HGVFs-PFG)algorithm is proposed to control fixed-wing UAVs to follow a generated guidance path and oriented target curves in three-dimensional space,which can be any combination of straight lines,arcs,and helixes as motion primitives.The algorithm aids the creation of vector fields(VFs)for these motion primitives as well as the design of an effective switching strategy to ensure that only one VF is activated at any time to ensure that the complex paths are followed completely.The strategies designed in earlier studies have flaws that prevent the UAV from following arcs that make its turning angle too large.The proposed switching strategy solves this problem by introducing the concept of the virtual way-points.Finally,the performance of the HGVFs-PFG algorithm is verified using a reducedorder autopilot and four representative simulation scenarios.The simulation considers the constraints of the aircraft,and its results indicate that the algorithm performs well in following both lateral and longitudinal control,particularly for curved paths.In general,the proposed technical method is practical and competitive.

Automated measurement of three-dimensional cerebral cortical thickness in Alzheimer’s patients using localized gradient vector trajectory in fuzzy membership maps

Our purpose in this study was to develop an automated method for measuring three-dimensional (3D) cerebral cortical thicknesses in patients with Alzheimer’s disease (AD) using magnetic resonance (MR) images. Our proposed method consists of mainly three steps. First, a brain parenchymal region was segmented based on brain model matching. Second, a 3D fuzzy membership map for a cerebral cortical region was created by applying a fuzzy c-means (FCM) clustering algorithm to T1-weighted MR images. Third, cerebral cortical thickness was three- dimensionally measured on each cortical surface voxel by using a localized gradient vector trajectory in a fuzzy membership map. Spherical models with 3 mm artificial cortical regions, which were produced using three noise levels of 2%, 5%, and 10%, were employed to evaluate the proposed method. We also applied the proposed method to T1-weighted images obtained from 20 cases, i.e., 10 clinically diagnosed AD cases and 10 clinically normal (CN) subjects. The thicknesses of the 3 mm artificial cortical regions for spherical models with noise levels of 2%, 5%, and 10% were measured by the proposed method as 2.953 ± 0.342, 2.953 ± 0.342 and 2.952 ± 0.343 mm, respectively. Thus the mean thicknesses for the entire cerebral lobar region were 3.1 ± 0.4 mm for AD patients and 3.3 ± 0.4 mm for CN subjects, respectively (p < 0.05). The proposed method could be feasible for measuring the 3D cerebral cortical thickness on individual cortical surface voxels as an atrophy feature in AD.

Corner-Based Image Alignment using Pyramid Structure with Gradient Vector Similarity

This paper presents a corner-based image alignment algorithm based on the procedures of corner-based template matching and geometric parameter estimation. This algorithm consists of two stages: 1) training phase, and 2) matching phase. In the training phase, a corner detection algorithm is used to extract the corners. These corners are then used to build the pyramid images. In the matching phase, the corners are obtained using the same corner detection algorithm. The similarity measure is then determined by the differences of gradient vector between the corners obtained in the template image and the inspection image, respectively. A parabolic function is further applied to evaluate the geometric relationship between the template and the inspection images. Results show that the corner-based template matching outperforms the original edge-based template matching in efficiency, and both of them are robust against non-liner light changes. The accuracy and precision of the corner-based image alignment are competitive to that of edge-based image alignment under the same environment. In practice, the proposed algorithm demonstrates its precision, efficiency and robustness in image alignment for real world applications.

Advection-Enhanced Gradient Vector Flow for Active-Contour Image Segmentation

In this paper,we propose a new gradient vector flow model with advection enhancement,called advection-enhanced gradient vector flow,for calculating the external force employed in the active-contour image segmentation.The proposed model is mainly inspired by the functional derivative of an adaptive total variation regularizer whose minimizer is expected to be able to effectively preserve the desired object boundary.More specifically,by incorporating an additional advection term into the usual gradient vector flow model,the resulting external force can much better help the active contour to recover missing edges,to converge to a narrow and deep concavity,and to preserve weak edges.Numerical experiments are performed to demonstrate the high performance of the newly proposed model.

Centerline Extraction for Image Segmentation Using Gradient and Direction Vector Flow Active Contours

In this paper, we propose a fast centerline extraction method to be used for gradient and direction vector flow of active contours. The gradient and direction vector flow is a recently reported active contour model capable of significantly improving the image segmentation performance especially for complex object shape, by seamlessly integrating gradient vector flow and prior directional information. Since the prior directional information is provided by manual line drawing, it can be inconvenient for inexperienced users who might have difficulty in finding the best place to draw the directional lines to achieve the best segmentation performance. This paper describes a method to overcome this problem by automatically extracting centerlines to guide the users for providing the right directional information. Experimental results on synthetic and real images demonstrate the feasibility of the proposed method.

GRIN(Gradient Index)介质中的Maxwell方程组与光线光学

利用坡印廷矢量(Poynting vector)的方向就是光线轨迹曲线的切线方向,推出程函方程(Eikonal equation)的矢量式.经分析发现此式包含了光的粒子性与光的波动性因素,光线的传播规律还受介质折射率函数的制约.再由程函方程进一步推得光线方程,并给出了应用实例.

Two Methods to Solve the Ionospheric Electron Concentration Horizontal Gradient at Chongqing

The electron concentration horizontal gradient vector of the ionosphere and its south-north and east-west components over Chongqing station are analyzed and calculated, using the first approximation, time correlation and space correlation and another approach introduced. And then, the validity of the two methods is analyzed and compared.

基于Gradient Boosting的车载LiDAR点云分类

车载LiDAR点云中包含地面、建筑物、行道树、路灯等丰富地物类别,自动对这些不同类别点云进行分类,对点云中目标的识别、提取及重建都具有重要意义。本文提出了一种基于Gradient Boosting的自动分类方法。该方法首先对车载激光点云进行数据预处理,然后计算点云的协方差矩阵、密度比、高程相关特征、局部平面特征、投影特征等,再计算点云特征直方图与垂直分布直方图,采用K-means方法对这两者分别进行聚类,并将其聚类类别值也作为特征,从而构建出20维的点云特征向量,应用Gradient Boosting分类方法进行自动分类。为了验证本文方法的有效性,从某城镇场景的车载激光点云数据中选取部分代表区域共144W点作为训练数据集,然后选取另一较大区域的点云共312W点作为测试数据集。使用训练好的分类器对测试数据集进行分类,分类结果总体准确率达到了93.38%,耗时631s,说明此分类方法具有较高的分类准确率,同时也具备较高的效率。

Conservative Vector Fields and the Intersect Rule

This paper covers the concept of a conservative vector field, and its application in vector physics and Newtonian mechanics. Conservative vector fields are defined as the gradient of a scalar-valued potential function. Gradient fields are irrotational, as in the curl in all conservative vector fields is zero, by Clairaut’s Theorem. Additionally, line integrals in conservative vector fields are path-independent, and line integrals over closed paths are always equal to zero, properties proved by the Gradient Theorem of multivariable calculus. Gradient fields represent conservative forces, and the associated potential function is analogous to potential energy associated with said conservative forces. The Intersect Rule provides a new, unique shortcut for determining if a vector field is conservative and deriving potential functions, by treating the indefinite integral as a set of infinitely many functions which satisfy the integral.

Polar 3D Transformation of the Full Gradient of Attractive Potential

The method of 3D polar transformation of full gravity potential gradient vectors is based on the geometric properties of the crossing points of complete gradient of the potential to localize the source region that causes the observed anomaly. The cross-points—poles—are defined for rectangular polygons of different sizes where the full gradient vector is defined at every vertex. The polygon size range could be specified. The set of poles, positive and negative, is then represented on the 3D chart in the form of clusters of dots or cubes and can be considered as a model image of the sources, intended for visual analysis and further interpretation.

计及采样扰动抑制的电压源逆变器三矢量无模型预测电流控制方法

针对传统电压源逆变器无模型预测电流控制(model-free predictive current control,MFPCC)方法存在电流纹波大、电流梯度更新停滞以及预测性能易受采样扰动影响的问题。该文提出一种计及采样扰动的三矢量MFPCC方法。在一个控制周期应用3个基本矢量,并根据价值函数计算矢量作用时间,降低了输出电流纹波;其次,通过建立不同矢量作用下的电流梯度方程组,实现电流梯度数据的实时更新,消除了停滞现象;再次,分析采样扰动对MFPCC的影响,采用扩张状态观测器估计采样扰动以补偿预测电流控制,抑制其对输出电流的影响。最后,通过仿真和实验,对所提方法的有效性进行了验证。

Tunable optical gradient trap by radial varying polarization Bessel-Gauss beam

Optical tweezers play an important role in many domains, especially in life science. And optical gradient force is necessary for constructing optical tweezers. In this paper, the optical gradient force in the focal region of radial varying polarization Bessel- Gauss beam is investigated numerically by means of vector diffraction theory. Results show that the beam parameter and vary rate parameter that indicates the change speed of polarization rotation angle affect the optical gradient force pattern very considerably, and some novel force distributions may come into being, such as multiple force minimums, force ring, and force crust. Therefore, the focusing of radial varying polarization Bessel-Gauss beam can be used to construct optical traps.

A Fast Algorithm for Training Large Scale Support Vector Machines

The manuscript presents an augmented Lagrangian—fast projected gradient method (ALFPGM) with an improved scheme of working set selection, pWSS, a decomposition based algorithm for training support vector classification machines (SVM). The manuscript describes the ALFPGM algorithm, provides numerical results for training SVM on large data sets, and compares the training times of ALFPGM and Sequential Minimal Minimization algorithms (SMO) from Scikit-learn library. The numerical results demonstrate that ALFPGM with the improved working selection scheme is capable of training SVM with tens of thousands of training examples in a fraction of the training time of some widely adopted SVM tools.

一种基于指令MKS的自动向量化代价模型

自动向量化代价模型是编译器进行自动向量化优化时的重要组成部分,其作用是评估代码在应用向量化转换后能否获得性能提升。当代价模型不准确时,编译器会应用负收益的向量化转换,从而降低程序的执行效率。针对GCC编译器默认代价模型的不精确问题,以Intel Xeon Silver 4214R CPU为平台,提出了一种基于指令MKS的自动向量化代价模型。该模型充分考虑了指令的机器模式、运算类型以及运算强度等,并使用梯度下降算法自动搜索不同指令类型的近似代价。在SPEC2006以及SPEC2017上进行了单线程测试,实验结果表明,该模型能够减少收益评估错误的情况。与默认代价模型生成的向量程序相比,GCC编译器添加MKS代价模型后,在SPEC2006课题上最高获得了4.72%的提速,在SPEC2017课题上最高获得了7.08%的提速。

结合SVM与XGBoost的链式多路径覆盖测试用例生成

机器学习方法可很好地与软件测试相结合,增强测试效果,但少有学者将其运用于测试数据生成方面.为进一步提高测试数据生成效率,提出一种结合SVM(support vector machine)和XGBoost(extreme gradient boosting)的链式模型,并基于此模型借助遗传算法实现多路径测试数据生成.首先,利用一定样本训练若干个用于预测路径节点状态的子模型(SVM和XGBoost),通过子模型的预测精度值筛选最优子模型,并根据路径节点顺序将其依次链接,形成一个链式模型C-SVMXGBoost(chained SVM and XGBoost).在利用遗传算法生成测试用例时,使用训练好的链式模型代替插桩法获取测试数据覆盖路径(预测路径),寻找预测路径与目标路径相似的路径集,对存在相似路径集的预测路径进行插桩验证,获取精确路径,计算适应度值.在交叉变异过程中引入样本集中路径层级深度较大的优秀测试用例进行重用,生成覆盖目标路径的测试数据.最后,保留进化生成中产生的适应度较高的个体,更新链式模型C-SVMXGBoost,进一步提高测试效率.实验表明,C-SVMXGBoost较其他各对比链式模型更适合解决路径预测问题,可提高测试效率.并且通过与已有经典方法相比,所提方法在覆盖率上提高可达15%,平均进化代数也有所降低,在较大规模程序上其降低百分比可达65%.

基于三维激光点云数据的电力电缆绝缘缺陷识别

针对电力电缆绝缘缺陷识别精准度较低的问题,提出一种基于三维激光点云数据的电力电缆绝缘缺陷识别方法。利用剪裁法增强原始三维激光点云数据,通过区域生长法与最小二乘法完整获取电缆绝缘材料的三维结构面信息。借助Canny边缘检测方法求解电缆绝缘表面缺陷与内部缺陷边缘信息,自动识别出电力电缆绝缘的缺陷位置及缺陷类别。结果表明,所提方法可以精准识别电缆绝缘表面的划痕缺陷、电缆外屏蔽表面起泡和孔洞缺陷,识别耗时短,鲁棒性较优,具有较高实际应用价值。

应用机器学习算法模型预测兴安落叶松地上生物量

为了准确预测兴安落叶松地上生物量,以小兴安岭201株兴安落叶松地上生物量作为研究对象,以胸径(D)和树高(H)为变量,构建随机森林(RF)、人工神经网络(ANN)、支持向量回归(SVR)和梯度提升回归树(GBRT)等4种机器学习模型,并将机器学习算法的预测结果与传统二元生物量模型的预测结果进行对比分析。结果表明:对比传统生物量模型,4种机器学习算法的拟合效果与检验精度均有了大幅度提高。模型拟合精度由高到低的顺序为随机森林、梯度提升回归树、人工神经网络、支持向量回归、传统生物量模型;RF模型在各模型中的拟合精度最高,相对于传统生物量模型,RF模型的确定系数(R~2)提升了3.72%,均方根误差(R_(MSE))降低了44.47%,平均绝对误差(M_(AE))降低了42.81%,相对误差绝对值(M_(PB))降低了42.80%,赤池信息准则值降低了18.17%。模型检验精度由高到低的顺序为随机森林、人工神经网络、梯度提升回归树、支持向量回归、传统生物量模型;RF模型在各模型中的预测精度最高,与传统生物量模型相比,RF模型的确定系数(R~2)提升了1.08%,均方根误差(R_(MSE))降低了10.95%,平均绝对误差(M_(AE))降低了10.34%,相对误差绝对值(M_(PB))降低了10.34%,赤池信息准则值降低了5.20%。因此,相对于传统生物量模型,4种机器学习算法模型均可以提高兴安落叶松地上生物量的预测精度,RF模型的预测精度最高。

基于FPGA并行实现SVM训练的可重构计算系统

针对支持向量机在处理大规模数据集时所面临的计算复杂度高和训练时间长的问题,设计了一种基于FPGA并行实现支持向量机训练的可重构计算系统,并分析了不同量化方式下的硬件资源消耗与加速性能。通过采用随机梯度下降法训练支持向量机,使得需要求解的维度与样本的维度相关联,相较于传统的基于二次规划的求解方法可以显著降低计算复杂性。同时,利用基于FPGA的可重构硬件平台设计了专用并行计算结构以加速支持向量机的训练过程。对设计的完整系统进行了软硬件联合仿真,在4个公共数据集上的仿真结果表明,整体模型预测准确率达到90%以上;在训练阶段,相较于采用相同算法的软件实现,所提出的浮点数表示下硬件实现的单个样本处理时间至少减少了2个数量级;定点数表示下硬件实现的单个样本处理时间最大减小了3个数量级;与基于二次规划问题求解的硬件实现相比,单个样本处理速度最快提升了394倍。

基于混合多机器学习算法的燃料电池性能退化预测框架

质子交换膜燃料电池(Proton exchange membrane fuel cell,PEMFC)是当下极具发展潜力的绿色发电装置,对其性能退化状态进行精准预测有助于促进电池健康管理,优化电池成本效益。为提高预测模型拟合度与精确度,提出用鹈鹕算法(Pelican optimization algorithm,POA)优化最小二乘支持向量机(Least squares support vector machine,LSSVM)的PEMFC性能退化预测模型。采用小波阈值去噪(Wavelet threshold denoising,WTD)与轻量级梯度提升机(Light gradient boosting machine,LGBM)进行数据预处理,以摒弃噪声与小关联度输入变量对预测的干扰。通过提供不同工况下电池运行数据和设立对比实验。结果表明,本模型的精度与稳定性优于其他模型,且占用资源较少,均方根误差保持在0.1%内,平均绝对百分比误差小于0.05%,性能优秀。