A Correntropy-based Affine Iterative Closest Point Algorithm for Robust Point Set Registration

The iterative closest point(ICP)algorithm has the advantages of high accuracy and fast speed for point set registration,but it performs poorly when the point set has a large number of noisy outliers.To solve this problem,we propose a new affine registration algorithm based on correntropy which works well in the affine registration of point sets with outliers.Firstly,we substitute the traditional measure of least squares with a maximum correntropy criterion to build a new registration model,which can avoid the influence of outliers.To maximize the objective function,we then propose a robust affine ICP algorithm.At each iteration of this new algorithm,we set up the index mapping of two point sets according to the known transformation,and then compute the closed-form solution of the new transformation according to the known index mapping.Similar to the traditional ICP algorithm,our algorithm converges to a local maximum monotonously for any given initial value.Finally,the robustness and high efficiency of affine ICP algorithm based on correntropy are demonstrated by 2D and 3D point set registration experiments.

A Review of Point Feature Based Medical Image Registration

Point features, as the basis of lines, surfaces, and bodies, are commonly used in medical image registration. To obtain an elegant spatial transformation of extracted feature points, many point set matching algorithms(PMs) have been developed to match two point sets by optimizing multifarious distance functions. There are ample reviews related to medical image registration and PMs which summarize their basic principles and main algorithms separately. However,to data, detailed summary of PMs used in medical image registration in different clinical environments has not been published. In this paper, we provide a comprehensive review of the existing key techniques of the PMs applied to medical image registration according to the basic principles and clinical applications. As the core technique of the PMs, geometric transformation models are elaborated in this paper, demonstrating the mechanism of point set registration. We also focus on the clinical applications of the PMs and propose a practical classification method according to their applications in different clinical surgeries. The aim of this paper is to provide a summary of pointfeaturebased methods used in medical image registration and to guide doctors or researchers interested in this field to choose appropriate techniques in their research.

Multi-view ladar data registration in obscure environment

Multi-view laser radar （ladar） data registration in obscure environments is an important research field of obscured target detection from air to ground. There are few overlap regions of the observational data in different views because of the occluder, so the multi-view data registration is rather difficult. Through indepth analyses of the typical methods and problems, it is obtained that the sequence registration is more appropriate, but needs to improve the registration accuracy. On this basis, a multi-view data registration algorithm based on aggregating the adjacent frames, which are already registered, is proposed. It increases the overlap region between the pending registration frames by aggregation and further improves the registration accuracy. The experiment results show that the proposed algorithm can effectively register the multi-view ladar data in the obscure environment, and it also has a greater robustness and a higher registration accuracy compared with the sequence registration under the condition of equivalent operating efficiency.

Coherent Point Drift Registration Combined with Image Feature and its Application

A key step of constructing active appearance model is requiring a set of appropriate training shapes with well-defined correspondences.In this paper,we introduce a novel point correspondence method(FB-CPD),which can improve the accuracy of coherent point drift(CPD) by using the information of image feature.The objective function of the proposed method is defined by both of geometric spatial information and image feature information,and the origin Gaussian mixture model in CPD is modified according to the image feature of points.FB-CPD is tested on the 3D prostate and liver point sets through the simulation experiments.The registration error can be reduced efficiently by FB-CPD.Moreover,the active appearance model constructed by FB-CPD can obtain fine segmentation in 3D CT prostate image.Compared with the original CPD,the overlap ratio of voxels was improved from 88.7% to 90.2% by FB-CPD.

Image Registration Technique Based on a Fast CPD Algorithm

The Coherent Point Drift (CPD) algorithm which based on Gauss Mixture Model is a robust point set registration algorithm. However, the selection of robustness weight which used to describe the noise may directly affect the point set registration efficiency. For resolving the problem, this paper presents a CPD registration algorithm which based on distance threshold constraint. Before the point set registration, the inaccurate template point set by resampling become the initial point set of point set matching, in order to eliminate some points that the distance to target point set is too close and too far in the inaccurate template point set, and set the weights of robustness as . In the simulation experiments, we make two group experiments: the first group is the registration of the inaccurate template point set and the accurate target point set, while the second group is the registration of the accurate template point set and the accurate target point set. The results of comparison show that our method can solve the problem of selection for the weight. And it improves the speed and precision of the original CPD registration.

一种基于人体轮廓形变场的双流网络步态识别方法

步态识别易受相机视角、服装和携带物等外界因素影响而性能下降。为此,该文将非刚性点集配准引入步态识别,利用相邻步态帧之间的形变场表征行走过程中人体轮廓发生的位移量,从而提升对人体形态变化的动态感知能力。在此基础上,该文提出一种基于人体轮廓形变场的双流卷积神经网络GaitDef,该网络模型由形变场和步态剪影两路特征提取分支构成。针对形变场数据的稀疏性设计多尺度特征提取模块,以获取形变场的多层次空间结构信息。针对步态剪影提出动态差异捕捉模块和上下文信息增强模块,以捕捉动态区域的变化特性和利用上下文信息增强步态表征能力。双分支网络的输出特征经过特征融合得到最终的步态表示。大量实验结果表明了该文方法的有效性,在CASIA-B和CCPG数据集上,该文方法的平均Rank-1准确率分别能达到93.5%和68.3%。

基于Lie群表示的保体积2D-3D点集配准算法

2D-3D点集配准的目标是寻找三维原始点集与二维目标投影点集之间的对应关系和最优变换.为了给出配准问题的解析解,避免投影引起的体积退化,提出基于Lie群表示的保体积2D-3D点集配准算法.首先,考虑投影矩阵和旋转矩阵的非交换性,引入Lie群表示,将配准问题形式化为一个Lie群优化问题.利用局部线性化方法,将Lie群优化问题转化为一个可计算的二次规划问题.其次,为了避免体积退化,考虑约束变换后的三维点集的投影与二维目标点集的投影具有相同的体积.为便于计算,引入Jensen-Bregman LogDet散度作为保体积正则项,将计算点集的体积差异转化为计算协方差矩阵之间的差异.然后,通过交替求解对应关系和最优变换,形成完整且可解的迭代策略.最后,在两个经典数据集上进行对比实验和消融实验,验证了该算法的精确性和有效性.

泰勒级数准则的船体分段测量数据配准解析解

为提高船体分段测量数据与CAD模型间的配准定位效率,本文给出了一个基于泰勒级数准则的点集配准解析解,直接求解配准变换参数。通过分析最小二乘解析解的求解特点,考虑到配准变换参数的齐次表示问题,引入特征因子构建新的齐次变换矩阵,建立配准齐次表示模型。利用泰勒级数展开齐次变换矩阵的最近正交矩阵,得到齐次变换矩阵的无限次估计解。在此基础上,根据配准齐次表示模型的最大化,计算旋转矩阵的近似解析最优解。以理论模型和甲板分段实测数据为例子验证。本文结果表明了提出算法的有效性及实用性,为船体分段快速测量分析提供算法依据。

面向盾牌形貌重建的双层采样点云粗配准方法

针对盾牌表面形貌三维重建过程中点云粗配准精度低和耗时长的问题,文中基于传统RANSAC(Random Sample Consensus)点云粗配准算法框架提出了一种双层采样点云粗配准算法。在第1层每次迭代中,采用单点采样进行刚性约束来减小对应集的大小,将此对应集作为第2层的“内点候选集”。算法第2层执行两个点的连续随机采样以及计算各自的最小模型,通过迭代最大化一致性集合并利用最小二乘法求解出最佳刚体变换矩阵。在实验阶段,采用6组降采样程度不同的盾牌点云,使用RANSAC算法和本文算法进行点云粗配准对比实验。实验结果表明,所提算法在粗配准速度和粗配准精度两方面都优于RANSAC算法,配准速度约为RANSAC算法的两倍,均方根误差为10-3 mm,比RANSAC算法的精度提高近400倍。

Novel 3D point set registration method based on regionalized Gaussian process map reconstruction

Point set registration has been a topic of significant research interest in the field of mobile intelligent unmanned systems.In this paper,we present a novel approach for a three-dimensional scan-to-map point set registration.Using Gaussian process(GP)regression,we propose a new type of map representation,based on a regionalized GP map reconstruction algorithm.We combine the predictions and the test locations derived from the GP as the predictive points.In our approach,the correspondence relationships between predictive point pairs are set up naturally,and a rigid transformation is calculated iteratively.The proposed method is implemented and tested on three standard point set datasets.Experimental results show that our method achieves stable performance with regard to accuracy and efficiency,on a par with two standard methods,the iterative closest point algorithm and the normal distribution transform.Our mapping method also provides a compact point-cloud-like map and exhibits low memory consumption.

基于WHI及ICP的点云配准算法

针对两步点云配准中精度差、计算效率低、易受噪声干扰的问题,提出一种基于WHI特征描述符结合改进的ICP点云配准算法。首先,对大数据量的点云通过ISS算法提取特征点集作为配准点云;然后,计算特征点云的WHI特征描述符,利用随机采样一致性算法完成粗配准;最后,基于安德森加速迭代ICP算法对粗配准点云进行精确配准。通过多组点云数据集对所提算法进行验证,实验表明,该算法配准精度高、速度快,在含有噪声数据集的优势更明显。在不同的点云模型下,所提算法的配准效率提高2倍以上,在噪声环境下具有一定的鲁棒性。

基于加权运动平均的多视角点云鲁棒配准方法

本文提出一种基于加权运动平均的多视角点云配准方法,有效提高了多视角点云配准的效率和精度.首先,通过双视角点云配准算法得到点集的相对运动集;提出一种分层渐进式点云配准初值确定方法,该方法计算运动关系图中三视匹配元一致性度量,逐步挑选最为可靠的匹配元进行顶点初值化与顶点扩展,有效避免了因所选初值与真值偏差较大导致加权运动平均效果不佳的问题;依据上述方法确定配准初值,并根据该初值判断双视角匹配结果的可靠性,以剔除相对运动集内包含的外点,最后将剔除外点后的原始相对运动集的内点子集作为加权运动平均算法输入的相对运动集.以斯坦福公开数据集为对象开展了对比实验,基于双视角配准算法分别获得了4个测试点集在30%、25%和20%双视角点云重叠率阈值下的双视角配准结果,共形成12组相对运动集.其中,相对运动集误差随重叠率阈值的下降逐渐增加.所提方法在12组运动集上均获得了最佳的配准结果,证明了本文所提出方法在提高加权运动平均方法计算精度与效率方面的有效性.此外,当运动集包含大量外点时,采用本文所提方法仍能获得较为准确的配准结果,证明了本文所提方法的强鲁棒性.

叶片六点优化定位规划技术研究

叶片具有结构复杂、自由曲面多等特点。为解决叶片密集测量数据与其CAD模型的预配准定位问题,提出六点优化定位算法进行预配准。结合六点定位原理与预配准分析技术,在榫头基准面上建立原始坐标系,对六点的主次方向进行定位误差分析配准运算,减小每个测点的误差,使CAD模型与理论模型达到吻合状态,最后完成6个自由度的点分布,得到最终的叶片模型预配准控制点集。避免了ICP算法对大量点云数据处理的繁琐计算,实现了批量叶片在六点优化定位后的密集测点优化预配准,且定位精度均在0.02 mm左右。六点优化定位与典型六点定位进行配准结果对比分析表明,叶片六点优化定位算法预配准精度均在0.02 mm左右,满足叶片模型预配准精度要求。

基于佳点集人工鱼群的点云配准算法

针对点云配准迭代最近点(iterative closest point,ICP)算法对点云的初始位置姿态有较高的要求且易陷入局部最优的问题,文章提出一种基于佳点集人工鱼群的点云配准算法。首先采用佳点集方法对人工鱼群初始化,解决人工鱼群因初始种群分布不均而陷入局部最优的问题,并通过下采样与三维尺度不变特征变换(3D scale invariant feature transform,3D SIFT)特征点提取简化点云;然后采用快速点特征直方图(fast point feature histogram,FPFH)特征描述解求点云间的对应点对并剔除错误对应点对,通过佳点集人工鱼群算法寻优刚性变换的6个参数完成粗配准;最后使用ICP算法完成精配准。实验选取斯坦福大学提供的Bunny、Dragon和Happy Buddha 3组测试数据集进行配准;结果表明,该文算法收敛速度快,能为ICP算法提供良好的初始位姿避免其陷入局部最优。

一种大尺寸视觉测量系统标定方法

针对相距较远的基坐标系与视觉传感器坐标系的标定问题,提出一种高精度标定方法,通过激光跟踪仪建立中间坐标系,基于最小二乘点集配准误差标定激光跟踪仪坐标系与视觉传感器坐标系的关系,采用最小二乘平均刚体变换矩阵标定激光跟踪仪坐标系与基坐标系之间的关系,最后通过坐标变换建立测量基坐标系与视觉传感器坐标系之间的坐标关系,实现大尺寸工件空间相位角度的精确测量.仿真结果证明了最小二乘点集配准方法求解坐标系关系的有效性,同时验证了最小二乘点集配准方法的配准误差与配准点集大小、传感器精度、传感器的位姿关系有关.试验结果表明,该标定方法能够准确可靠地测量大尺寸工件空间角度,测量标准差达到0.007 5°,完全满足工艺要求.

基于点云稀疏语义特征的智能网联汽车协同感知配准算法

针对道路场景下多智能网联汽车协同感知问题,本文提出一种基于点云稀疏语义特征的智能网联汽车协同感知配准算法。所提算法旨在通过点云集成配准扩展智能网联汽车感知范围,进而实现智能网联汽车协同感知。首先,在道路语义特征基础上进行几何特征提取进而得到点云稀疏语义特征。其次,计算道路语义特征点云间的角度偏差以提供配准初值,并将点云语义信息作为配准约束条件实现全局语义集成配准。实验表明所提算法有效扩大了多智能网联汽车协同感知范围,提高了多点云集成配准的精度与鲁棒性。与当前主流算法JRMPC相比,本文所提算法配准精度提高了2.45%。

叶片边缘破损曲面再生修复技术研究

发动机叶片长期在高温、高压和高速下工作,工作过程中有异物进入,不可避免地会发生损坏。对于弯扭叶片边缘破损曲面的修复,本文采用基于邻近破损域未破损区域的完整叶片截面线与破损域叶片截面线的完好部分配准,获得边缘破损处的叶片截面线点云特征,实现破损曲面再生重构。同时,采用迭代最近点((Iterative closest point,ICP))配准算法,克服了搜索最近点集耗时较长的缺点。本文提出的基于建立辅助圆邻域的最近点集对搜索方法,大大降低了配准时间成本。最后对破损域再生曲面进行了光顺性分析,验证本文所提出的边缘破损曲面再生修复方法的可行性。

迭代最近点算法研究进展

曲面配准问题是三维计算机视觉中的一个极其重要的问题,作为解决深度图像的配准问题的一个应用较为广泛的算法,ICP算法得到了研究者的关注,本文以一种全新的思路从控制点的选取、特征度量、空间搜索、点对权重和刚体变换的求解等5个方面对深度图像配准的ICP算法的各种改进和优化进行了分类和总结。并提出了该算法尚需研究和改进的方向。

基于CBCT的体表标记点辅助配准方法在体部肿瘤放疗摆位中的应用研究

目的:应用基于CBCT(cone beam computed tomography)的体表标记点辅助图像配准方法,提高肿瘤放射治疗摆位验证的工作效率。方法:选取胸部、腹部、盆腔共30例肿瘤放疗患者,定位及治疗摆位时体表粘贴标记点。CBCT扫描验证时,医师、主管技师、技师先进行常规图像配准,后采用体表标记点辅助图像配准,以医师常规图像配准为基础分别进行比较。结果:常规图像配准时,主管技师对比医师在胸部、腹部及盆腔部位的X、Y、Z轴上的差异均无统计学意义;技师对比医师在腹部的X、Y、Z轴上及胸部、盆腔部位的X、Y轴上的差异无统计学意义,在胸部及盆腔部位的Z轴上P值分别为0.006,0.014,有显著意义。标记点辅助图像配准时,医师、主管技师及技师对比医师常规图像配准在胸部、腹部及盆腔部位的X、Y、Z轴上的差异均无统计学意义。图像配准的误差发生率腹部大于胸部及盆腔。医师、主管技师、技师分别与自己对比在胸部、腹部及盆腔部位的图像配准完成时间上P值近似于0,有显著意义。结论:以体表标记点辅助图像配准优于常规图像配准,效率高。

空间非合作目标的多视角点云配准算法研究

为了提高相邻视角间稀疏扫描点云数据配准的速度和精度,实现多视角点云精确配准,提出一种基于KD-Tree点云均匀采样简化算法,并且对传统四点算法(4-Points Congruent Sets Algorithm,4PCS)中的阈值参数进行了统一,确定了各误差阈值参数和点云密度之间的关系,通过基于姿态校正的方法有效解决了对称视角点云引起的误配准问题。仿真结果表明,该方法能够快速、有效地实现卫星稀疏点云的配准。