移动相机下基于三维背景估计的运动目标检测

室内环境中的运动目标检测是计算机视觉领域的研究热点,而移动相机造成的动态背景是运动目标检测的难点。本文提出一种基于同步定位与地图创建(ORB-SLAM)三维背景估计的运动目标检测算法,首先使用移动相机遍历整个室内环境,采用ORB-SLAM技术建立当前全局环境的三维背景特征点云模型;然后基于局部视频建立局部三维特征点云,根据定位信息将当前局部三维特征点云与环境三维背景特征点云进行嵌入,基于环境背景信息,采用三维均值漂移(3DMS)算法,对局部三维特征点云进行前景特征点提取;运用深度卷积神经网络,对提取的前景特征点所在候选区域进行运动目标确认。通过多个室内场景的实际实验进行验证,结果表明本文方法具有较高的运动目标检测准确率和召回率,提出的运动目标检测算法充分利用了三维背景信息,采用深度卷积神经网络进行确认,有效地改善了检测的准确性和鲁棒性。     

基于深度学习的三维点云分析方法研究进展

点云是目前自动驾驶、机器人、遥感、增强现实(AR)、虚拟现实(VR)、电力、建筑等领域最常用的三维数据处理形式,深 度学习方法能够处理大型数据,且可自主提取特征,因此点云深度学习方法已逐渐成为研究热点。 本文综述了近十年来基于深 度学习的三维点云分析方法的研究进展。 首先给出了三维点云深度学习的相关概念;然后针对点云目标检测与跟踪、分类分 割、配准和匹配以及拼接这 4 种任务,分别阐述了相应的深度学习方法的原理,分析并比较了各自的优缺点;随后整理了 18 种 点云数据集和 4 种点云分析任务的性能评价指标,并给出了性能对比结果;最后总结了点云分析方法目前存在的问题,对进一 步的研究工作进行了展望。     

3D obstacle detection of indoor mobile robots by floor detection and rejection

Obstacle detection is essential for mobile robots to avoid collision with obstacles.Mobile robots usually operate in indoor environments,where they encounter various kinds of obstacles; however,2D range sensor can sense obstacles only in 2D plane.In contrast,by using 3D range sensor,it is possible to detect ground and aerial obstacles that 2D range sensor cannot sense.In this paper,we present a 3D obstacle detection method that will help overcome the limitations of 2D range sensor with regard to obstacle detection.The indoor environment typically consists of a flat floor.The position of the floor can be determined by estimating the plane using the least squares method.Having determined the position of the floor,the points of obstacles can be known by rejecting the points of the floor.In the experimental section,we show the results of this approach using a Kinect sensor.     

基于直线光流场的三维检测新方法

提出了基于直线光流场的三维检测计算方法。在透视投影下,将表示二维直线的参数对时间的导数定义成直线光流场,详细推导出基于直线光流场检测三维刚体运动参数的模型。使用最小二乘法,求出刚体的三维旋转运动参数、平移运动参数和空间直线坐标,从而实现了刚体的三维检测。该方法利用单目图像序列中连续3幅图像的2对直线光流,能够检测出物体的运动参数和结构参数,有助于全面检测和识别物体。多组实验表明该方法是稳定的,具有较好的鲁棒性能。     

飞机三维姿态测量的角平分线方向向量法

基于交会测量提出了一种飞机三维姿态测量的新方法—角平分线方向向量法。对交会图像进行边缘检测和特征线提取,获得飞机两机翼边缘特征线上适当两点的坐标,经两次坐标转换到特定坐标系下;联合经纬仪参数及跟踪的俯仰、方位角,解算机翼边缘特征线在特定坐标系下的方向向量,利用飞机结构的对称性,在机翼特征线平面求解其两条角平分线方向向量;重构飞机中轴线及其同面垂线的方向向量;据此计算出飞机的俯仰、偏航和滚动三维姿态。利用两台光电经纬仪交会跟踪飞机的眼镜蛇机动飞行表明提出的方法稳定、实用,只要图像上机翼边缘线成像清晰,测量的姿态角误差不会超过1°。此方法不仅可为靶场光测设备跟踪图像的处理提供一种新途径,而且可为其他运动目标的姿态测量提供借鉴。     

基于变分光流的三维运动检测与稠密结构重建

提出了一种基于变分理论的由稠密光流检测图像中目标物体三维运动与结构的直接方法。首先在透视投影模型下将目标物体的三维运动参数与图像光流联系起来,构建了一个未知量是三维运动参数的能量函数,然后使用变分方法直接求出图像中目标运动物体的三维运动参数,最后由求出的三维运动参数恢复物体表面的三维稠密结构及图像光流。该方法在对目标物体进行三维运动检测时省略了"光流计算"这一中间过程,避免了图像光流计算给三维运动检测结果带来的误差影响和时间消耗。多组实验证明该算法具有较高的计算精度、较好的鲁棒性和较快的计算效率。     

基于三维检测的升船机螺母柱加工工艺

螺母柱是平衡重式垂直升船机安全保障系统的关键设备,材料特殊,精度要求高,加工难度大.文中采用特殊的工艺措施以及专用刀具来保证其制造精度,并在大型工件加工过程中创造性的引入三维检测方法,对零件加工过程及温度变化造成的影响进行控制,有效地改善了螺母柱高精加工时采用“试切法”带来的缺点,充分考虑经济性,最终满足实际工程要求.     

矩形脉冲涡流传感器的三维磁场量与缺陷定量评估

脉冲涡流是一种有效的电磁无损检测技术。基于脉冲涡流检测原理设计了脉冲涡流检测系统,并对矩形脉冲涡流传感器的三维磁场量进行了研究。分别在传感器的不同走向下,使用三维检测传感器获得了三维磁场量的Bx、By与Bz曲线。在传感器的不同走向下,提出使用三维蝶形图对缺陷进行检测识别。对三维信号进行特征分析后,分别在不同走向下,通过Bz曲线评估缺陷的深度。通过By与Bz曲线的特征可以测量缺陷的长度。为进一步实现飞机机身缺陷的成像检测提供了有价值的参考。     

Microscopic brain tumor detection and classification using 3D CNN and feature selection architecture

Brain tumor is one of the most dreadful natures of cancer and caused a huge number of deaths among kids and adults from the past few years. According to WHO standard, the 700,000 humans are being with a brain tumor and around 86,000 are diagnosed since 2019. While the total number of deaths due to brain tumors is 16,830 since 2019 and the average survival rate is 35%. Therefore, automated techniques are needed to grade brain tumors precisely from MRI scans. In this work, a new deep learning‐based method is proposed for microscopic brain tumor detection and tumor type classification. A 3D convolutional neural network (CNN) architecture is designed at the first step to extract brain tumor and extracted tumors are passed to a pretrained CNN model for feature extraction. The extracted features are transferred to the correlation‐based selection method and as the output, the best features are selected. These selected features are validated through feed‐forward neural network for final classification. Three BraTS datasets 2015, 2017, and 2018 are utilized for experiments, validation, and accomplished an accuracy of 98.32, 96.97, and 92.67%, respectively. A comparison with existing techniques shows the proposed design yields comparable accuracy.     

3D and 2D structural characterization of 1D Al/Al2O3 biphasic nanostructures

1D Al/Al₂O₃ nanostructures have been synthesized by chemical vapour deposition (CVD) of the molecular precursor [^tBuOAlH₂]₂. The deposited nanostructures grow chaotically on the substrate forming a layer with a high porosity (80%). Depending on the deposition time, diverse nanostructured surfaces with different distribution densities were achieved. A three‐dimensional (3D) reconstruction has been evaluated for every nanostructure density using the Focus Ion Beam (FIB) tomography technique and reconstruction software tools. Several structural parameters such as porosity, Euler number, geometrical tortuosity and aspect ratio have been quantified through the analysis with specified software of the reconstructions. Additionally roughness of the prepared surfaces has been characterized at micro‐ and nanoscale using profilometry and AFM techniques, respectively. While high aspects ratio around 20–30 indicates a strong anisotropy in the structure, high porosity values (around 80%) is observed as a consequence of highly tangled geometry of such 1D nanostructures.     

高光谱成像结合人工神经网络无损检测桃的硬度

为无损检测桃的内部品质,提出了结合高光谱成像技术和人工神经网络无损检测桃硬度的方法。采集了摘后贮藏了12d的140个桃在900~1 700nm的高光谱图像,以每个桃高光谱图像中40pixel×40pixel的感兴趣区域的平均光谱作为桃的原始反射光谱;利用Savitzky-Golay平滑和标准正态变量变换对光谱进行预处理;基于x-y共生距离算法划分样本,得到校正集样本105个和预测集样本35个。利用连续投影算法、无信息变量消除法和正自适应加权算法从全光谱的216个波长中分别提取了12个、103个和22个特征波长;分别建立了基于全光谱和提取的特征波长预测桃硬度的支持向量机模型和BP网络模型。结果表明,基于全光谱建立的BP网络模型具有最好的预测性能,其预测相关系数为0.856,预测均方根误差为0.931。本研究为基于桃内部品质的工业化分级提供了基础。     

复合材料车削表面2D分析和3D分析的对比研究

这里对碳/碳复合材料车削圆柱表面进行了二维轮廓分析和三维形貌分析的对比研究,分析计算表明三维形貌分析方法更能够真实的反映复合材料的表面质量。     

3D elemental and structural analysis of biological specimens using electrons and ions

We demonstrate the utility of focused ion beam scanning electron microscopy combined with energy dispersive x-ray spectrometry for 3D morphological and elemental correlative analysis of subcellular features. Although recent advances in super-resolution light microscopy techniques and traditional transmission electron microscopy methods can provide cellular imaging at a wide range of length scales, simultaneous 3D morphological and elemental imaging of cellular features at nanometre scale can only be achieved with techniques such as focused ion beam scanning electron microscopy with energy dispersive x-ray spectrometry capability. We demonstrate the technique by analysing the 3D silicon cell wall structure of a marine diatom, Thalassiosira pseudonana. This study also highlights the limitations of the technique in its current state and suggests several possible improvements needed for the routine use of the technique for biological specimens.     

叉排翅片传热特性的二维与三维数值分析

换热器是一种广泛应用于工业生产的设备,而叉排翅片(OSF)因具有良好的换热性能而被广泛应用于换热器。目前基于二维(2D)与三维(3D)数值方法,针对OSF流动传热特性已展开了大量研究。为了进一步加深对实际边界条件下的OSF传热特性的研究,通对层流和过渡流下的OSF进行2D与3D的非定常数值分析,找出二者在预测换热性能方面的差异并分析导致差异的影响因素,从而寻找一种利用2D或3D数值方法模拟OSF流动传热特性的有效途径,研究结果表明:在层流区3D效应较小,采用2D数值方法可更为有效地对理想的OSF传热特性进行定性和定量的预测;在过渡流区3D效应加强,由于OSF底板边界层对非定常流的抑制作用,2D数值分析方法不能准确地预测OSF的传热性能。     

直齿圆柱齿轮的三维建模及有限元分析

以Pro/E为开发平台,Pro/Program为开发工具,对标准直齿圆柱齿轮的参数化建模进行了详细的介绍.只要按提示输入标准直齿圆柱齿轮的5个基本参数,就可以得到精确的直齿齿轮的三维实体.同时利用ANSYS有限元分析软件的特点,可以在不脱离Pro/E和ANSYS这两个软件的情况下对直齿圆柱齿轮进行有限元分析.     

Experimental methods for local-scale characterization of hydro-morphodynamic dam breach processes. Breach detection, 3D reconstruction,flow kinematics and spatial surface velocimetry

A set of tests on failure by overtopping of earth dams was performed under hydraulic and geotechnical controlled conditions in a medium-scale facility located at the Portuguese National Laboratory for Civil Engineering. The flow over the breaching dam is three-dimensional and exhibits several length scales. Instrumentation must be able to capture this complexity and scale amplitude and must be non-intrusive. This work presents a summary of the instrumentation layout as well as its main components and measuring methods that were used and/or developed to address these issues. The emphasis is placed on post-processing analysis of digital imagery, for which there is detailed exposition of the adopted methods and an evaluation of its suitability for its intended purpose.     

圆弧金刚石砂轮三维几何形貌的在位检测和误差评价

针对非球面光学元件加工对圆弧金刚石砂轮形状误差测量的需求,提出了砂轮三维几何形貌在位检测与误差评价方法。建立了砂轮外圆面螺旋扫描轨迹测量数学模型,利用位移传感器获取了砂轮表面轮廓数据;对得到的数据匀滑滤波后沿圆周展开并进行插值处理,得到砂轮三维几何形貌。然后,根据非球面平行磨削加工特点,提出评价圆弧砂轮形状精度的指标。通过提取三维几何形貌轴截面轮廓,进行最小二乘圆弧拟合得到不同相位处的圆弧半径与圆心坐标,并由误差分离获得砂轮表面圆弧的圆度误差、圆周跳动误差及轮廓圆心轴向偏差。最后,对非球面加工圆弧金刚石砂轮进行检测,获得了砂轮的三维几何形貌以及多个关键尺寸及其误差数据:即圆弧金刚石砂轮的平均圆弧半径为55.442 3mm,半径波动极差为0.16mm,中央±8mm环带内圆弧的圆度误差约为5μm,圆周跳动误差约为2μm,截面轮廓圆心轴向位置相对偏差为0.008mm。根据检测结果,进行了大口径复杂非球面磨削实验,得到的元件面形P-V值为4.62μm,RMS值优于0.7μm,满足工程的实际需求。     

基于数字化3D技术的股骨假体再设计与激光选区熔化制造

研究了基于数字化3D技术的全膝置换股骨假体再设计并通过激光选区熔化(SLM)技术直接制造了股骨假体,以满足医学上对全膝置换股骨假体的高适配性要求。对一名患者的全膝关节CT连续断层图像提取股骨3D模型,根据骨科医生手术规划进行了数字化3D解剖与测量,并据此对目前商业化的假体进行了重新设计。然后,利用SLM技术直接制造了再设计完成的3D股骨假体模型,并讨论了制造工艺参数、机械性能、空间优化摆放位置以及成型精度等关键技术。实验结果显示:依据患者股骨远端解剖参数可完成股骨假体的3D模型再设计并可利用SLM技术直接制造出股骨假体,单个股骨假体成型时间为5.2h,成型精度标准偏差为0.030mm,成型致密度达到99.02%;热处理后成型性能优于美国实验材料学会(ASTM)F75的铸造标准。得到的结果表明该项技术可以快速制造完成患者所需要的股骨假体,且成型性能优良。