关于P—point与■_(1)-scale的个数

本文考察P-Point与_1-scale的个数,证明了下面结论:(ⅰ)如果CH或MA成立,则至少存在2^(_0)多个P-point,(ⅱ)“存在2^(_1)多个_1-scale”是协调的。

Formula for calculating spatial similarity degrees between point clouds on multi-scale maps taking map scale change as the only independent variable

The degree of spatial similarity plays an important role in map generalization, yet there has been no quantitative research into it. To fill this gap, this study first defines map scale change and spatial similarity degree/relation in multi-scale map spaces and then proposes a model for calculating the degree of spatial similarity between a point cloud at one scale and its gener- alized counterpart at another scale. After validation, the new model features 16 points with map scale change as the x coordinate and the degree of spatial similarity as the y coordinate. Finally, using an application for curve fitting, the model achieves an empirical formula that can calculate the degree of spatial similarity using map scale change as the sole independent variable, and vice versa. This formula can be used to automate algorithms for point feature generalization and to determine when to terminate them during the generalization.

多尺度特征融合对比学习结合PointMetaBase的点云分割

点云场景边界的精确分割对提升三维点云整体分割精度和小目标的分割精度都非常重要。针对现有三维点云语义分割算法中存在的对点云场景边界分割不准确的问题,设计了一种新的多尺度特征融合对比学习方法,并将其集成到PointMetaBase网络上,提出了MFFCL-PMB三维点云语义分割网络。该网络将解码器各层的输入和输出平行多路径地输入多尺度特征提取网络,其次将多尺度特征进行拼接融合,最后由边界搜索模块确定点云场景边界点,在边界点对应的多尺度融合特征上进行对比学习,使网络学习到使特征区分度更大的权重。MFFCL-PMB在数据集S3DIS的测试集区域5上的mIoU为70.9%,相比原始的PointMetaBase,在边界上的mIoU增加了1.4%,内部区域mIoU增加了1.2%,总体mIoU增加了1.2%,边界mIoU增量比内部区域增加的mIoU更大,说明MFFCL-PMB有效增强了PointMetaBase在边界上的分割表现并提高了网络在点云总体上的语义分割性能。

基于特征融合的改进型PointPillar点云目标检测

针对PointPillar在自动驾驶道路场景下对点云稀疏小目标检测效果差的问题,通过引入一种多尺度特征融合策略和注意力机制,提出一种点云目标检测网络Pillar-FFNet。针对网络中的特征提取问题,设计了一种基于残差结构的主干网络;针对馈入检测头的特征图没有充分利用高层特征的语义信息和低层特征的空间信息的问题,设计了一种简单有效的多尺度特征融合策略;针对主干网络提取的特征图中信息冗余的问题,提出了一种卷积注意力机制。为验证所提算法的性能,在KITTI和DAIR-V2X-I数据集上进行实验。实验结果表明,所提出的算法在KITTI数据集上与PointPillar相比,汽车、行人和骑行者的平均精度最大提高分别为0.84%,2.13%和4.02%;在DAIR-V2X-I数据集上与PointPillar相比,汽车、行人和骑行者的平均精度最大提高分别为0.33%,2.09%和4.71%,由此证明了所提方法对点云稀疏小目标检测的有效性。

PointPCA:一种三维空间点云的特征提取算法

点云是一个庞大点的集合而且拥有重要的几何结构。由于其庞大的数据量,不可避免地就会在某些区域内出现一些相似点,这就使得在进行特征提取时提取到一些重复的信息,造成计算冗余,降低训练的准确率。针对上述问题,提出了一种新的神经网络——PointPCA,可以有效地解决上述问题;在PointPCA中,总共分为三个模块:a)采样模块,提出了一种average point sampling(APS)采样方法,可以有效地规避一些相似的点,得到一组近似代表这组点云的新的点集;b)特征提取模块,采用分组中的思想,对这组新的点的集合进行多尺度空间特征提取;c)拼接模块,将每一尺度提取的特征向量拼接到一起组合为一个特征向量。经过实验表明,PointPCA比PointNet在准确率方面提升了4.6%,比PointNet++提升了1.1%;而且在mIoU评估测试中也有不错的效果。

Effects of slip condition on MHD stagnation-point flow over a power-law stretching sheet

The steady two-dimensional magnetohydrodynamic stagnation flow towards a nonlinear stretching surface is studied. The no-slip condition on the solid boundary is replaced with a partial slip condition. A scaling group transformation is used to get the invariants. Using the invariants, a third-order ordinary differential equation corresponding to the momentum is obtained. An analytical solution is obtained in a series form using a homotopy analysis method. Reliability and efficiency of series solutions are shown by the good agreement with numerical results presented in the literature. The effects of the slip parameter, the magnetic field parameter, the velocity ratio parameter, the suction velocity parameter, and the power law exponent on the flow are investigated. The results show that the velocity and shear stress profiles are greatly influenced by these parameters.

An Improved Affine-Scaling Interior Point Algorithm for Linear Programming

In this paper, an Improved Affine-Scaling Interior Point Algorithm for Linear Programming has been proposed. Computational results of selected practical problems affirming the proposed algorithm have been provided. The proposed algorithm is accurate, faster and therefore reduces the number of iterations required to obtain an optimal solution of a given Linear Programming problem as compared to the already existing Affine-Scaling Interior Point Algorithm. The algorithm can be very useful for development of faster software packages for solving linear programming problems using the interior-point methods.

Existence of Positive Solutions for Higher Order Boundary Value Problem on Time Scales

In this paper, we investigate the existence of positive solutions of a class higher order boundary value problems on time scales. The class of boundary value problems educes a four-point （or three-point or two-point） boundary value problems, for which some similar results are established. Our approach relies on the Krasnosel＇skii fixed point theorem. The result of this paper is new and extends previously known results.

Attractive fixed-point solution study of shell model for homogeneous isotropic turbulence

The attractive fixed-point solution of a nonlinear cascade model is stud- ied for the homogeneous isotropic turbulence containing a parameter C, introduced by Desnyansky and Novikov. With a traditional constant positive external force added on the first shell equation, it can be found that the attractive fixed-point solution of the model depends on both the parameter C and the external force. Thus, an explicit force is introduced to remove the effects of the external force on the attractive fixed-point solu- tion. F^arthermore, two groups of attractive fixed-point solutions are derived theoretically and studied numerically. One of the groups has the same scaling behavior of the velocity in the whole inertial range and agrees well with those observed by Bell and Nelkin for the nonnegative parameters. The other is found to have different scaling behaviors of the velocity at the odd and even number shells for the negative parameters. This special characteristic may be used to study the anomalous scaling behavior of the turbulence.

Positive Solutions for First-order PBVPs on Time Scales

In this paper, the author studies the following nonlinear dynamic equation {x△(t) = r(t)x(σ(t)) + f(t, x(σ(t))), t ∈ [0, T ], x(0) = x(σ(T )). By applying and improving the generalized form of Leggett-Williams fixed point theorem, sufficient conditions are established for the existence of positive solutions.

A class of boundary value problems for third-order differential equation with a turning point

A class of boundary value problems for a third-order differential equation with a turning point is considered. Using the method of multiple scales and others, the uniformly valid asymptotic expansion of solution for the boundary value problem is constructed.

ASIC Design of Floating-Point FFT Processor

An application specific integrated circuit (ASIC) design of a 1024 points floating-point fast Fourier transform(FFT) processor is presented. It can satisfy the requirement of high accuracy FFT result in related fields. Several novel design techniques for floating-point adder and multiplier are introduced in detail to enhance the speed of the system. At the same time, the power consumption is decreased. The hardware area is effectively reduced as an improved butterfly processor is developed. There is a substantial increase in the performance of the design since a pipelined architecture is adopted, and very large scale integrated (VLSI) is easy to realize due to the regularity. A result of validation using field programmable gate array (FPGA) is shown at the end. When the system clock is set to 50 MHz, 204.8 μs is needed to complete the operation of FFT computation.

Spectral matching algorithm based on nonsubsampled contourlet transform and scale-invariant feature transform

A new spectral matching algorithm is proposed by us- ing nonsubsampled contourlet transform and scale-invariant fea- ture transform. The nonsubsampled contourlet transform is used to decompose an image into a low frequency image and several high frequency images, and the scale-invariant feature transform is employed to extract feature points from the low frequency im- age. A proximity matrix is constructed for the feature points of two related images. By singular value decomposition of the proximity matrix, a matching matrix （or matching result） reflecting the match- ing degree among feature points is obtained. Experimental results indicate that the proposed algorithm can reduce time complexity and possess a higher accuracy.

基于PointPillars改进的点云目标检测算法

现有的点云目标检测方法虽然层出不穷,但主要以提升检测精度为主,检测效率较低,很难满足实际应用时的实时性需求。PointPillars[1]方法通过一种创新的点云特征处理方法,明显提升了点云目标检测的速度,为工业界的实际应用提供了可能性,但其精度相比速度略有欠缺。论文基于多尺度特征融合和3D注意力机制,对PointPillars方法的结构进行改进,提高了原方法对于多尺度目标的检测精度,弥补了该方法相比主流方法在检测精度上的不足。改进后的方法兼具较高的检测精度和较快的检测速度,具有一定的工程应用价值。

应用4-scale模型对人工针叶林郁闭度的反演

利用4-scale模型模拟不同叶面积指数、不同样地树木结构参数下的背景可视概率、冠层反射率,建立查找表。利用遥感数据,以像元尺度ILA值作为控制变量,根据冠层反射率数据查找对应ILA下的背景可视概率,进而反演出冠层郁闭度。最后采用鱼眼相机与样点法实测的郁闭度做模型的检验。结果表明:4-scale模型能很好地反演人工针叶林冠层郁闭度,与其他3种测量郁闭度方法相比,4-scale模型反演郁闭度与鱼眼照片提取的冠层郁闭度有很好的线性关系,决定系数R^2=0.628、均方根误差ERMS=0.0476,与样点法、样线法计算的郁闭度R^2与ERMS分别为0.5591、0.0518和0.5395、0.0529。考虑树干对鱼眼照片提取冠层郁闭度的影响后R^2为0.6847;对于人工落叶松林,R^2由0.4226提升到0.6153;对于人工油松林,R^2由0.7646变为0.7561。4-scale模型针能很好的反演人工针叶林冠层郁闭度,同时也证明4-scale模型反演冠层郁闭度与鱼眼照片提取的冠层郁闭度之间存在良好的关系。

基于RS-Conv的多尺度神经网络LiDAR点云断裂带提取方法

断裂带与地震、滑坡等自然灾害的发生有着密切关系,精准提取断裂带不仅可为地震断层的定量化研究提供指导,还可为地震灾害风险评估及防震减灾决策的制定提供科学依据。针对现有方法中LiDAR点云断裂带提取不完整、连续性差及错误率高等问题,文中提出了一种基于RS-Conv的多尺度神经网络LiDAR点云断裂带提取方法,以便更好地解决复杂地形区域的断裂带自动提取问题。该方法首先构建不同空间尺度的邻域点集,从而更全面地考察点云的局部几何结构特征。考虑到RS-Conv算子能够很好地表征中心点与邻域点的空间关系,文中以RS-Conv算子作为卷积模块构建了多尺度神经网络模型,以提取出LiDAR点云不同尺度的深层次特征,对其进行堆叠并输入到全连接层,以完成对断裂带点的提取。最后,在ISPRS点云数据集、川滇点云数据集和鲜水河数据集上对文中所述方法与张量分解方法和Deep Neural Networks(DNN)方法进行了对比实验,结果表明,文中方法的分类精度最高,分类总误差最低仅为0.3%,较其他方法降低了0.91%~2.79%,证实了该方法在点云断裂带提取方面的优越性。

基于SIFT特征点提取的ICP配准算法

为解决传统迭代最近点(ICP)算法对点云配准的起始点对选择不佳而导致配准时间长、效率低的问题,提出一种基于尺度不变特征变换(SIFT)特征点提取的ICP点云配准算法(ST-ICP)。首先使用SIFT算法进行原始点云与目标点云的SIFT特征点提取,根据提取特征点完成快速点特征直方图(FPFH)特征运算,通过采样一致性初始配准算法(SAC-IA)搜索对应点对、求解变换矩阵,再进一步运用ICP算法进行点云精细配准。实验结果表明:与ICP算法相比较,ST-ICP算法的配准误差在迭代次数为5次时减小了1.019 cm,迭代次数为10次时减小了0.443 cm;在配准误差达到10^(-2) cm级别时,ST-ICP算法所用时间比传统ICP算法减少了12.829 s。ST-ICP算法优化了对应点对的选择,提升了配准精度和配准效率。

Periodic Solutions for a Kind of Second-Order Neutral Differential Equation on Time Scales

In this paper,the existence of periodic solutions and nonnegative periodic solutions for a kind of second-order neutral differential equation on time scales is considered.The partial increment is a fixed constant if the time scale is the real number set and is a multiple of the periodic of the time scale if the time scale is not the real number set.By means of a fixed point theorem due to Krasnoselskii,some sufficient conditions are obtained.