Model-based orientation-independent 3-D machine vision techniques

Orientation-dependent techniques for the identification of a three-dimensional (3-D) object by a machine vision system are represented in parts. In the first part, the data consist of intensity images of polyhedral objects obtained by a single camera, while in the second part, the data consist of range images of curved objects obtained by a laser scanner. In both cases, the attributed graph representation of the 3-D object surface is used to drive the respective algorithm. In this representation, a graph node represents a surface path and a link, the adjacency between two patches. The attributes assigned to nodes are moment invariants of the corresponding face for polyhedral objects. For range images, the Gaussian curvature is used as a segmentation criterion for providing symbolic shape attributes. Identification is achieved by an efficient graph-matching algorithm used to match the graph obtained from the data to a subgraph of one of the model graphs stored in the computer memory     

基于单数码相机自由拍摄的空间点定位

 提出并实现了一种空间点定位方法。该方法仅用一台数码相机围绕目标进行多角度自由拍摄获得多帧图像，采用一套具有身份唯一性的标记点实现多图像间点的稳定匹配；在此基础上根据多视图几何约束，精确计算各次拍摄时的相机姿态和位置；进而求解目标点的三维位置信息，并对结果进行优化。实验表明空间点定位的相对误差优于0.03%。该方法所用设备十分简单，操作方式非常灵活，适用于大型物体多视角测量的拼合、某些情况下的装配检查等，也可直接用于较规则构型物体的三维数据获取与CAD模型重建。     

Extrinsic calibration of a laser displacement sensor in a non-contact coordinate measuring machine

In order to implement 3D scanning of those complicated parts such as blades in the aviation field,a non-contact optical measuring system is established in the paper,which integrates a laser displacement sensor,a probe head,the frame of a coordinate measuring machine (CMM),etc.As the output of the laser sensor directly obtained possesses the 1D length of the laser beam,it needs to determine the unit direction vector of the laser beam denoted as (l,m,n) by calibration so as to convert the 1D values into 3D coordinates of target points.Therefore,an extrinsic calibration method based on a standard sphere is proposed to accomplish this task in the paper.During the calibration procedure,the laser sensor moves along with the motion of the CMM and gathers the required data on the spherical surface.Then,both the output of the laser sensor and the grating readings of the CMM are substituted into the constraint equation of the spherical surface,in which an over-determined nonlinear equation group containing unknown parameters is established.For the purpose of solving the equation group,a method based on non-linear least squares optimization is put forward.Finally,the system after calibration is utilized to measure the diameter of a metallic sphere 10 times from different orientations to verify the calibration accuracy.In the experiment,the errors between the measured results and the true values are all smaller than 0.03 mm,which manifests the validity and practicality of the extrinsic calibration method presented in the paper.     

Extension of continuous scanning laser Doppler vibrometry measurement for complex structures with curved surfaces

The continuous Scanning Laser Doppler Vibrometry (SLDV) developed on the base of the galvanometer scanner system has made it possible to quickly obtain the full field vibration responses within a rectangular area of the structure. In this paper, an arbitrary continuous scanning path generating method for Continuous Scanning Laser Doppler Vibometry (CSLDV) is further put forward in order to allow the CSLDV suitable for testing structures featured by complex shapes not just for regular areas. In the first step, the relationship between position of laser spot and the driving voltages of galvanometer scanner system has been described by a mathematical modeling. Then, a novel arbitrary scanning path generating strategy based on CSLDV is presented by deforming a normalization rectangular scanning path to an arbitrary continuous scanning path. The mapping relation between the normalization rectangular scanning path and arbitrary continuous scanning path is established using the reference points. In the second step, a compressor blade with curved surface was taken as an example for modal test using the proposed method. At the same time, a validated experiment was performed in SLDV. The results show the mode shapes derived from the extended CSLDV are in agreement with those from SLDV and the Modal Assurance Criterion (MAC) between the two are all greater than 0.96. They also demonstrate the feasibility and effectiveness of the proposed method for CSLDV test and show strong potential on further practical engineering applications.     

逆向工程中曲面重构算法研究与实现

给出了基于坐标测量机测得的海量数据重新构造样板物体的数字化表面模型的算法流程。该算法首先利用部分测量数据构造张量积的B样条网格曲面,通过确定适当的检查点,并计算检查点到所构造的B样条网格曲面间的距离,从而控制重构曲面的构造精度。讨论了与文中算法相关的空间点到NURBS曲面间的距离计算等问题。给出了1个工程应用实例,表明该算法稳定可靠、效率较高。     

基于三维激光扫描的蒙皮对缝检测研究

针对传统蒙皮对缝检测方法误差大、效率低等问题以及二维激光对缝检测的缺陷,采用了一种基于三维激光扫描蒙皮对缝的检测方法并开发出相应软件。提出了蒙皮对缝三维点云数据的去噪、临界点识别及提取方法,根据蒙皮对缝间隙阶差对气动外形的影响,分析其实际几何结构,与理论结构对比,提炼出了计算蒙皮对缝间隙阶差的数学模型,并通过对试验模拟件对缝的检测,验证了其计算精度。结果表明:该检测方法可行,与二维激光检测相比精度和重复精度更高,间隙测量精度达到0.04mm,阶差测量精度达到0.02mm,可以满足蒙皮对缝检测需求,为分析间隙阶差对装配质量的影响提供数值依据。     

三维激光成像系统的点云模型真彩色处理方法

三维激光成像系统获得的点云数据精度高并且保有高精度的高程信息,然而特征信息不明显,没有保留全部的点和点的拓扑关系,无法直接判读地面物体;影像数据空间分辨率和精度低,并且缺少高程信息,然而其影像的特征连续,光谱信息丰富.因此,进行点云数据与影像数据融合方法的研究是目前三维激光成像系统研制的前沿性课题.针对点云数据与影像数据融合方法的研究,提出了一种三维激光成像系统点云模型真彩色处理的方法.实验结果显示,采用所提出的方法,不仅能够获得被扫描物体的位置信息和大小信息,而且能够得到被扫描物体的表面纹理细节信息.     

基于改进HOG特征的空间非合作目标检测

传统的非合作目标检测方法大都基于一定的匹配模板,这不仅需要预先指定先验信息,进而设计合适的检测模板,而且同一模板只能对具有相似形状的目标进行检测,不易直接用于检测形状未知的非合作目标。为降低检测过程中对目标形状等先验信息的要求,借鉴基于规范化梯度的物体区域估计方法,提出一种基于改进方向梯度直方图特征的目标检测方法,首先构建包含有自然图像和目标图像的训练数据集;然后提取标记区域的改进方向梯度直方图特征,以更好地保持局部特征的结构性,并根据级联支持向量机训练模型,从数据集中自动学习目标物体的判别特征;最后,将训练后的模型用于检测测试集图像中的目标。实验结果表明,算法在由4953幅和100幅图像构成的测试集中分别取得94.5%和94.2%的检测率,平均每幅图像的检测时间约为0.031 s,具有较低的时间开销,且对目标的旋转及光照变化具有一定的鲁棒性。     

激光制导武器光斑形状半实物仿真技术研究

在激光制导武器半实物仿真试验中，其中一个要求就是激光目标模拟器为导引头提供的光斑形状应与其在实战环境下“看到”的一致。根据激光制导武器制导原理，建立了实战条件下激光光斑形状模型，并设计了光斑形状半实物仿真系统。依据仿真相似性原理，建立了仿真条件下的光斑形状数学模型及激光目标模拟器发射的激光束散角数学模型，讨论了照射误差对光斑形状的影响。通过仿真分析证明这些模型是正确的。     

激光宏观结构化金刚石砂轮磨削氧化铝工艺

为了研究砂轮表面结构化对砂轮磨削性能的影响,利用脉冲激光对树脂结合剂金刚石砂轮进行了表面宏观结构化。采用6种不同类型的金刚石砂轮表面宏观结构进行了氧化铝的磨削实验,建立了激光宏观结构化金刚石砂轮的磨削力模型,比较了6种不同激光宏观结构化金刚石砂轮与非结构化砂轮在不同磨削参数下磨削力的差异,分析了砂轮制造后的表面形貌与结构化砂轮的磨损特性。实验结果表明,砂轮宏观结构化对磨削性能有很大影响,激光宏观结构化砂轮的磨削力可以减小2. 5%~24. 5%,砂轮结构化后的表面形貌出现石墨化现象;宏观结构化砂轮沟槽边缘磨损加剧,但沟槽磨损并没有明显加快宏观结构化砂轮的磨损。     

透射式激光等离子体微烧蚀推力性能研究

激光等离子体微推进技术经过近10年的发展,越来越受到业内人士的广泛关注。针对透射式激光等离子体微烧蚀物理过程,构建了理论分析模型,并进行了仿真。结合仿真结果,分析了激光脉宽和靶材的选取对烧蚀产生的推力和冲量的影响,提出了激光微烧蚀推进参数设计过程中,需要注意的几个问题。采用冲击摆系统测量激光微烧蚀靶材推力特性,对理论模型进行修正,与实验结果对比,模型与实验测量结果符合良好。     

Cleaning space debris with a space-based laser system

High-energy pulsed laser radiation may be the most feasible means to mitigate the threat of collision of a space station or other valuable space assets with orbital debris in the size range of 1–10 cm. Under laser irradiation, part of the debris material is ablated and provides an impulse to the debris particle. Proper direction of the impulse vector either deflects the object trajectory or forces the debris on a trajectory through the upper atmosphere, where it burns up. Most research concentrates on ground-based laser systems but pays little attention to space-based laser systems.There are drawbacks of a ground-based laser system in cleaning space debris. Therefore the placement of a laser system in space is proposed and investigated. Under assumed conditions,the elimination process of space debris is analyzed. Several factors such as laser repetition frequency, relative movement between the laser and debris, and inclination of debris particles which may exercise influence to the elimination effects are discussed. A project of a space-based laser system is proposed according to the numerical results of a computer study. The proposed laser system can eliminate debris of 1–10 cm and succeed in protecting a space station.     

激光脉冲法背温信号中噪声信号的滤除方法

针对激光脉冲法背温信号中存在的噪声干扰,通过频谱分析确定了噪声信号的频率范围,设计了相应的FIR数字滤波器,并对滤波后存在的残留噪声进行了递推平滑滤波.滤波前后的数据对比表明,该方法能有效滤波背温信号中的噪声.     

激光推进发射微小卫星弹道仿真与分析

建立了初步的激光大气传输模型和光船的三维运动模型,通过对单站大气盘旋入轨模式下的推进弹道仿真分析可知,单站推进模式下激光在稠密大气层中行程较大,激光传输衰减严重。针对上述问题,以激光推进发射微小卫星为应用目标,提出了基于过顶接力和等光程接力两种方式下三站入轨的推进模式,给出并分析了相应的仿真结果,并利用等光程接力方式分析了激光推进发射微小卫星的运载能力。结果表明,三站等光程接力方式可以有效的解决激光传输问题,减小了激光在稠密大气层中的传输距离和激光的传输损失,提高入轨质量。     

基于红外视觉/激光雷达融合的目标识别与定位方法

近年来,基于可见光图像的目标识别在无人车感知领域得到了广泛应用.然而,可见光图像目标识别无法应用于弱光和黑暗环境.针对于此,提出了一种基于红外视觉/激光雷达融合的目标识别与定位算法.首先,通过基于颜色迁移的数据增强训练方法,提高了红外目标识别算法的泛化性能.继而,提出了一种基于激光雷达修正的单目深度估计方法,通过视觉图像与激光雷达点云的数据融合,实现了基于稠密深度图对目标位置的估计,提高了对小目标的定位能力.试验结果表明,与传统Yolov3目标识别算法相比,该算法平均识别准确率可提升5.8％;此外,相对定位算法将包含小目标在内的物体相对定位精度提高了13.4％.     

基于深度学习的突防控制博弈对象匹配方法CSCD

针对基于博弈理论设计应对多枚拦截弹的协同突防控制方案时需要确定博弈对象的问题,提出了一种基于长短时记忆(LSTM)网络的拦截弹攻击对象匹配方法。基于传统防空导弹飞行时序与流程构建拦截弹飞行轨迹库,以轨迹库为训练样本对LSTM网络进行训练,并以此为基础构建航迹预测模型与对象匹配模型,实现对拦截弹攻击对象的识别。仿真结果表明,该方法能够有效识别拦截弹拦截目标,为后续的巡航弹突防研究提供支撑。     

提高深空异步激光测距精度方法研究

异步激光测距的作用距离远,有望实现对地球同步轨道以及深空航天器的精确测量。在对相关文献深入研究的基础上,全面分析了异步激光测距基本原理,对其钟差和测距公式进行了重新推导,提出了一种适用于深空异步激光测距的精确公式,并给出了迭代求解方法。通过理论仿真分析,在脉冲时间测量误差5ns时,对匀速飞行目标,该方法距离测量精度可达到5～6m,比其他方法高出2～3个数量级。     

基于逆向工程的发动机叶片数字化设计验证

对发动机叶片实体建模中的关键问题进行了分析,并采用关节臂激光扫描仪进行了装配模式下发动机叶片的逆向扫描,运用Geomagic Qualify软件实现了叶片型面三维误差比较以及叶片截面指标计算分析,为发动机叶片的数字化设计验证提供了参考。     

扩压器叶片形线逆命题设计的一种新方法

应用遗传算法结合神经网络技术通过对候选叶片表面速度分布与目标速度分布之间误差的极小化,求解给定表面速度分布下的扩压器叶片形线。叶片形线利用Bezier曲线参数化表示,候选叶片形线的流场分析由BP神经网完成。      

基于改进DeepSORT的视觉/激光雷达目标跟踪与定位方法CSCD

通过结合目标跟踪与相对定位,在对多帧检测目标进行关联与分析的同时,可以获取其三维信息。但当目标外观特征变换较大时,传统目标跟踪算法较易发生漏匹配或身份变换,而仅依靠对齐点云的相对定位算法较易出现定位失效的情况。针对以上问题,提出了一种基于改进DeepSORT的目标跟踪与定位方法在原始DeepSORT算法中加入基于位置约束的匹配,解决了因外观改变导致的漏匹配问题;在获取跟踪信息的基础上,设计了基于目标运动模型的相对定位方法,解决了图像中目标较小时相对定位不连续且定位精度较低的问题。试验结果表明,与传统DeepSORT算法相比,多目标跟踪准确度提高了5.9%;与仅依靠对齐点云的相对定位算法相比,定位精度提高了62.4%。