基于Point Transformer v2的点云枝叶分离方法研究

准确高效的点云枝叶分离对精确计算森林树木的垂直参数至关重要。然而,当前的研究方法计算成本高,且依赖先验知识导致泛化能力不足。针对以上问题,文章提出利用基于点特征的Transformer网络进行自动化的森林场景三维点云的枝叶分离研究。该方法使用Point Transformer v2网络,首先利用网格编码模块提取可学习的局部结构关系,保留点云的几何拓扑结构;其次使用分组注意力实现多通道联合学习,降低特征的冗余度,提高计算的效率;最后构建了基于点的Transformer网络实现高精度森林树木三维点云语义分割,降低了对于先验知识的需求。使用地基激光扫描仪获取的加拿大和芬兰7个不同树种样地的三维点云数据,进行枝叶分离实验和精度评价。实验结果表明,网络整体精度(OA)为94.42%,mIoU为78.89%,能够适应不同树种、不同点云密度的森林场景的枝叶分离。

Uncertainty Principle for the Quaternion Linear Canonical Transform in Terms of Covariance

An uncertainty principle(UP),which offers information about a signal and its Fourier transform in the time-frequency plane,is particularly powerful in mathematics,physics and signal processing community.Under the polar coordinate form of quaternion-valued signals,the UP of the two-sided quaternion linear canonical transform(QLCT)is strengthened in terms of covariance.The condition giving rise to the equal relation of the derived result is obtained as well.The novel UP with covariance can be regarded as one in a tighter form related to the QLCT.It states that the product of spreads of a quaternion-valued signal in the spatial domain and the QLCT domain is bounded by a larger lower bound.

基于S-YOLO V5和Vision Transformer的视频内容描述算法

视频内容描述的自动生成是结合计算机视觉和自然语言处理等相关技术提出的一种新型交叉学习任务。针对当前视频内容生成描述模型可读性不佳的问题,本研究提出一种基于S-YOLO V5和Vison Transformer(ViT)的视频内容描述算法。首先,基于神经网络模型KATNA提取关键帧,以最少帧数进行模型训练;其次,利用S-YOLO V5模型提取视频帧中的语义信息,并结合预训练ResNet101模型和预训练C3D模型提取视频静态视觉特征和动态视觉特征,并对两种模态特征进行融合;然后,基于ViT结构的强大长距离编码能力,构建模型编码器对融合特征进行长距离依赖编码;最后,将编码器的输出作为LSTM解码器的输入,依次输出预测词,生成最终的自然语言描述。通过在MSR-VTT数据集上进行测试,本研究模型的BLEU-4、METEOR、ROUGEL和CIDEr分别为42.9、28.8、62.4和51.4;在MSVD数据集上进行测试,本研究模型的BLEU-4、METEOR、ROUGEL和CIDEr分别为56.8、37.6、74.5以及98.5。与当前主流模型相比,本研究模型在多项评价指标上表现优异。

退火工艺对Ti-Ni-V形状记忆合金相变和形变行为的影响

利用差示扫描量热仪和拉伸试验研究了400℃×10~120 min和500℃×10~120 min退火态Ti-50.8Ni-0.5V合金的相变行为、拉伸性能和形状记忆行为。结果表明:随退火温度和时间变化,合金的相变类型、R相变温度T_(R)和热滞ΔT_(R)、马氏体相变温度T_(M)和热滞ΔT_(M)、抗拉强度R_(m)、伸长率A、应力诱发马氏体相变临界应力σ_(M)、加载-卸载残余应变ε_(R)和能耗W_(D)的变化规律如下:400和500℃退火态Ti-50.8Ni-0.5V合金皆发生B2→R→B19’/B19’→R→B2(B2-母相,CsCl型结构;R-R相,菱方结构;B19’-马氏体相,单斜结构)型相变;T_(R)^(400℃)>T_(R)^(500℃),T_(M)^(500℃)>T_(M)^(400℃),ΔT_(M)^(400℃)>ΔT_(M)^(500℃)≥ΔT_(R)^(400℃)≈ΔT_(R)^(500℃)≈4℃;R_(m)^(400℃)>R_(m)^(500℃),A^(500℃)>A^(400℃);400℃退火态合金的超弹性(SE)稳定性高于500℃退火态合金。随退火时间延长,T_(R)^(400℃)、T_(R)^(500℃)、T_(M)^(400℃)和T_(M)^(500℃)升高,ΔT_(M)^(400℃)和ΔT_(M)^(500℃)减小,ΔT_(R)^(400℃)和ΔT_(R)^(500℃)在3.5~4.6℃之间变化;400℃退火态合金保持SE,500℃退火态合金由SE向SE+SME(形状记忆效应)转变;σ_(M)^(400℃)和σ_(M)^(500℃)降低;W_(D)^(400℃)和W_(D)^(500℃)增加;ε_(R)^(500℃)先降后升;ε_(R)^(400℃)在0.64%~1.36%之间变化。要使Ti-50.8Ni-0.5V合金获得较高的T_(R),可对其进行400℃退火;要使该合金获得较高的T_(M),可对其进行500℃退火;要使该合金在室温下获得超弹性,可对其进行400℃×10~120 min或500℃×10~60 min退火。

改进Deeplabv3+的双注意力融合作物分类方法

近年来,卷积神经网络(convolutional neural networks,CNN)在农作物分类研究中不断取得新进展,但在建模长期依赖关系方面表现出一定的局限性,对农作物全局特征的捕获存在不足。针对以上问题,将Transformer引入Deeplab v3+模型,提出了一种用于无人机影像农作物分类的并行分支结构--DeepTrans(Deeplab v3+with Trans-former)模型。DeepTrans以一种并行的方式将Transformer和CNN结合在一起,利于全局特征与局部特征的有效捕获。通过引入Transformer来增强图像中信息的远距离依赖关系,提高了作物全局信息的提取能力;加入通道注意力机制和空间注意力机制加强Transformer对通道信息的敏感度及ASPP(atrous spatial pyramid pooling)对作物空间信息捕获能力。实验结果表明,DeepTrans模型在MIoU指标上可达0.812,相较于Deeplab v3+模型提高了3.9%,该模型在五类作物的分类中精度均有提升,对于容易错分的甘蔗、玉米和香蕉三种作物,其IoU分别提高了2.9%、4.7%、13%。由此可见,DeepTrans模型在农作物分类图像的内部填充和全局预测方面有着更好的分割效果,有助于更准确地监测农田作物的种植结构及规模。

基于改进YOLOv7的织物疵点小目标检测算法

针对织物疵点种类多、尺度变化大、小目标易漏检等问题,提出了一种基于YOLOv7的改进算法(YOLOv7⁃ESL)。首先,将注意力机制ECA⁃Net融入到Neck层中取代CBS模块,在少量增加计算成本的情况下显著提高了检测精度。其次,设计了专用探测头检测小目标,充分利用网络的浅层特征信息,使模型能够有效地检测多尺度的目标。最后,在特征加强部分增加Swin Transformer V2 Block,能够捕捉全局和局部之间的丰富关系,提高模型检测小目标缺陷的能力。试验结果表明:YOLOv7⁃ESL算法精确率为97.7%,召回率为90.3%,平均精度均值为94.9%,FPS为29.9帧/s;与原始YOLOv7模型相比,分别提高了3.8个百分点、3.4个百分点、3.3个百分点、3.1帧/s,可满足纺织工业领域的应用要求。

Fe-V和Ni-V二元体系热力学优化

采用CALPHAD(calculation of phase diagram)方法对Fe-V和Ni-V二元系进行了热力学优化,并结合第一性原理计算,利用(A,B)_(10)(A,B)_(4)(A,B)_(16)三亚点阵模型描述σ相,首次成功地描述了σ相的占位分数。使用优化的模型参数不仅可以描述Fe-V和Ni-V系热化学性质,而且可以很好地重现Fe-V和Ni-V系的相平衡关系,为高熵合金多组元热力学数据库的建立奠定基础。

V-5Cr-5Ti合金弹性和力学性质的第一性原理研究

V-5Cr-5Ti合金作为核聚变堆第一包层的主要候选结构材料之一,但对其力学性质的理论研究相对较少.采用随机固溶体模型,利用第一性原理方法计算出V-5Cr-5Ti合金的弹性常数、体模量、剪切模量、杨氏模量、泊松比和柯西压力等,并与计算出的纯钒的相关数值进行对比,结果表明V-5Cr-5Ti合金具有良好的塑性和强度,但其塑性要略低于纯钒的.并对加入氧原子后的V-5Cr-5Ti合金进行了相关计算,通过对比计算结果发现,由于氧原子的加入,使V-5Cr-5Ti合金的塑性和强度都出现了不同程度的降低.最后对V-5Cr-5Ti合金和纯钒的理论强度进行了计算,并绘制出两者的应力-应变关系图,通过对比再次验证了上面的结论.

The Fourier Transform and Principles of Quantum Mechanics

From the hypotheses that the position-representation of a physical state is the Fourier transform of its momentum-representation and that the timerepresentation is the inverse Fourier transform of its energy-representation, we are able to obtain the Planck relation E = hν , the de Broglie relation p = h /λ , the Dirac fundamental commutation relation, the Schr?dinger equations, the Heisenberg uncertainty principle in quantum mechanics, and the annihilation/creation of a photon from excitation/de-excitation of an atom following Bohr.

基于改进YOLO v5的苹果叶部病害检测

针对YOLO v5检测模型存在的漏检率和误检率问题,改进目标检测技术,提升苹果叶部病害早期发现及定位的准确性和速度,从而减少经济损失。先采用加权双向特征金字塔网络(BiFPN)特征融合方法,有效改善PANet对多尺度特征融合的不良影响,并引入Transformer机制,有效改进原始网络结构,使其专注于有用的信息,并增强特征融合效果。再使用ATCSP模块和自上而下的特征融合方法,增强模型对多尺度疾病的检测效果,并将强大的语义信息传达至模型底层,进一步提高检测精度。使用数据集由实验室采集的苹果树叶样本构成,包含3 331张标记图像的矩形位置来标记病害。由于图像亮度分布不均匀,采用直方图均衡化和改进的直方图均衡化处理,使图像对比度得到增强,大幅降低后续图像特征提取的计算量。在训练和测试过程中,还对原始病害图像进行旋转、随机亮度增强、随机色度增强、随机对比度增强和锐化等数据增强操作。结果表明,改进的YOLO v5检测算法可以显著提高苹果叶部病害检测的精度,对比原始算法,平均精度mAP@0.5提高20.8%。改进YOLO v5苹果叶部病害检测算法能够及时发现和定位苹果叶部病害,进而为深度学习技术在农业病害监测中的广泛应用提供技术支撑。

First-principles calculation of the lattice compressibility,elastic anisotropy and thermodynamic stability of V_2GeC

We investigate the elastic and the thermodynamic properties of nanolaminate V2GeC by using the ab initio pseudopotential total energy method. The axial compressibility shows that the c axis is always stiffer than the a axis. The elastic constant calculations demonstrate that the structural stability is within 0-800 GPa. The calculations of Young＇s and shear moduli reveal the softening behaviour at about 300 GPa. The Possion ratio makes a higher ionic or a weaker covalent contribution to intra-atomic bonding and the degree of ionicity increases with pressure. The relationship between brittleness and ductility shows that V2GeC is brittle in ambient conditions and the brittleness decreases and ductility increases with pressure. Moveover, we find that V2CeC is largely isotropic in compression and in shear, and the degree of isotropy decreases with pressure. The Griineisen parameter, the Debye temperature and the thermal expansion coefficient are also successfully obtained for the first time.

气体吸附对V掺杂石墨烯电子结构与光学性质的影响

基于密度泛函理论 (DFT )的广义梯度近似 (GGA ),采用第一性原理方法研究了气体分子吸附对V掺杂石墨烯的吸附能、电子结构与光学性质的影响.能带结构计算表明:吸附NO_(2) 分子的V掺杂石墨烯的带隙显著增加,从0 eV变为0.368 eV,由金属性转变为半导体特性,而吸附CO与NH_(3) 分子的V掺杂石墨烯的带隙则变化很小.三种吸附构型 (NO_(2) , CO, NH_(3) )的吸附能分别为-8.499 eV、-2.05 eV和-2.01 eV,说明V掺杂石墨烯对NO_(2) 气体分子吸附最强.进而计算了本征、V掺杂石墨烯及其吸附NO_(2) 分子的光学性质,结果表明:随着V掺杂与吸附NO_(2) 气体,石墨烯介电吸收峰值有所增大,介电峰位向低能量区域移动;本征石墨烯仅吸收紫外光,V掺杂石墨烯吸附NO_(2) 分子可以明显拓宽光吸收的光谱范围;掺杂与吸附使得石墨烯光电导率显著增强,能在红外与可见光区产生光电流.上述结果表明V掺杂石墨烯吸附NO_(2) 后材料电子结构和光学性质明显变化,能够作为光学气敏传感材料检测NO_(2) 气体.

RELATIVE PRINCIPLE COMPONENT AND RELATIVE PRINCIPLE COMPONENT ANALYSIS ALGORITHM

In this letter,the new concept of Relative Principle Component (RPC) and method of RPC Analysis (RPCA) are put forward. Meanwhile,the concepts such as Relative Transform (RT),Ro-tundity Scatter (RS) and so on are introduced. This new method can overcome some disadvantages of the classical Principle Component Analysis (PCA) when data are rotundity scatter. The RPC selected by RPCA are more representative,and their significance of geometry is more notable,so that the application of the new algorithm will be very extensive. The performance and effectiveness are simply demonstrated by the geometrical interpretation proposed.

基于V2C催化剂的混合电解质锂空气电池催化机理研究

氧还原反应(ORR)和析氧反应(OER)催化活性的提高可以显著提高混合电解质锂空气电池的性能。二维材料V2C(MXenes)具有丰富的组成、高比表面积和强稳定性等特点,受到广泛关注。本工作基于密度泛函理论(DFT)的第一性原理方法,研究了V2C的电子性质、碱性条件下氧气在V2C表面的最佳吸附位点和氧气在V2C表面的氧化还原过程,计算了V2C的自由能和过电势,最后将其与混合电解质锂空气电池常用的MnO2催化剂进行了比较。结果表明,V2C作为混合电解质锂空气电池催化剂时,比常用的MnO2催化剂的过电势更小,说明该催化剂可以提高电极的催化性能,是一种前景良好的ORR和OER双功能催化剂。

STUDY OF THE EQUIVALENT THEOREM OF GENERALIZED VARIATIONAL PRINCIPLES IN ELASTICITY

The relations of all generalized variational principles in elasticity are studied by employing the invariance theorem of field theory. The infinitesimal scale transformation in field theory was employed to investigate the equivalent theorem. Among the results found particularly interesting are those related to that all generalized variational principles in elasticity are equal to each other. Also studied result is that only two variables are independent in the functional and the stress-strain relation is the variational constraint condition for all generalized variational principles in elasticity. This work has proven again the conclusion of Prof. Chien Wei-zang.

FVIT-YOLO v8:基于多尺度融合注意机制的改进YOLO v8小目标检测

本文研究了遥感与无人机航拍图像中的小目标检测问题。由于这类图像存在目标尺度小、目标分布密集、背景复杂等特点,使得特征提取困难。目前针对小目标检测的算法,为了提升精度,大多忽略了参数量与推理速度的影响,这使得算法缺乏实用性。针对上述问题,本文提出了一种基于轻量化的多尺度融合注意机制的改进YOLO v8小目标检测算法。算法首先在YOLO v8的FPN结构中加入F算子,设计了多尺度特征的加权融合;然后在网络预测层剔除了P4、P5预测层,加入P2层用于小目标的预测;最后对轻量化自注意力机制进行图像输入网格化分割整合改进,并用它替换了FPN中的C2f模块,使得算法具有更好的全局感知能力,并大幅降低了参数量。与YOLO v8s相比,本文算法在DOTA数据集上的mAP提升了4.4%,网络参数量下降了52%,FPS达到了46帧/s。在VisDrone数据集中,本算法在精度上提升了8.2%。

Uncertainty Principles for the Generalized Fourier Transform Associated with Spherical Mean Operator

The aim of this paper is to establish an extension of qualitative and quantitative uncertainty principles for the Fourier transform connected with the spherical mean operator.

Heisenberg Uncertainty Principle for n-Dimen-sional Linear Canonical Transforms

The uncertainty principle proposed by German physicist Heisenberg in 1927 is a basic principle of quantum mechanics and signal processing.Since linear canonical transformation has been widely used in various fields of signal processing recently and Heisenberg uncertainty principle has been endowed with new expressive meaning in linear canonical transforms domain,in this manuscript,an improved Heisenberg uncertainty principle is obtained in linear canonical trans-forms domain.

Possible Martensitic Transformation in Heusler Alloy Pt_2MnSn from First Principles

Using density functional theory calculations, we investigate the tetragonal distortion, electronic structure and magnetic property of Pt2MnSn. The results indicate that, when the volume-conserving tetragonal distortion occurs, the energy minimum appears at c/a = 0.84, and the energy difference between the minimum and cubic phase is as high as 107 me V/f. u. Thus from the point of view of thermodynamics, martensitie transformation may occur in Pt2MnSn with decreasing the temperature. The electronic structure of its cubic and martensitic phases also approves this. Moreover, both the cubic and tetragonal phases of Pt2MnSn are ferromagnetic structures and their total magnetic moments are 4.26 μB and 4.12 μB, respectively.

Special Relativity with a Preferred Frame and the Relativity Principle

The purpose of the present study is to develop a counterpart of the special relativity theory that is consistent with the existence of a preferred frame but, like the standard relativity theory, is based on the relativity principle and the universality of the (two-way) speed of light. The synthesis of such seemingly incompatible concepts as the existence of preferred frame and the relativity principle is possible at the expense of the freedom in assigning the one-way speeds of light that exists in special relativity. In the framework developed, a degree of anisotropy of the one-way speed acquires meaning of a characteristic of the really existing anisotropy caused by motion of an inertial frame relative to the preferred frame. The anisotropic special relativity kinematics is developed based on the symmetry principles: 1) Space-time transformations between inertial frames leave the equation of anisotropic light propagation invariant and 2) a set of the transformations possesses a group structure. The Lie group theory apparatus is applied to define groups of transformations between inertial frames. Applying the transformations to the problem of calculating the CMB temperature distribution yields a relation in which the angular dependence coincides with that obtained on the basis of the standard relativity theory but the mean temperature is corrected by the terms second order in the observer velocity.