基于线性特征的鱼眼图像校正方法

针对鱼眼镜头采集的畸变图像,在棋盘格标定模板的基础上,提出一种有效的校正方法。该方法采用标准畸变模型,首先基于畸变图像的线性特征提出目标函数;然后利用SUSAN改进算法精确定位棋盘格角点坐标;对已得到的有限角点坐标进行畸变曲线拟合,在畸变曲线上提取所需数量的点;最后用渐近寻优算法计算最优畸变系数,使得目标函数最小。实验结果表明,该方法角点定位精确、算法时间复杂度低,能够有效校正鱼眼镜头造成的图像畸变。     

一种结构光三维扫描仪系统自动标定法

针对棋盘格特点设计新的棋盘格角点提取算法,利用标定棋盘格的特点实现了亚像素级的角点坐标自动化提取。实验结果证明,方法实现标定自动化的同时也保证了较高的标定精度和鲁棒性。     

基于对称性分析的棋盘图像角点检测方法

为实现摄像机标定过程中棋盘图像角点的自动检测,提出一种基于对称性分析的的角点检测方法.该方法主要包括图像特征点提取、初始角点检测和角点检测3部分.首先,根据棋盘图像角点与图像边缘的关系提取图像候选特征点并对其进行聚类.通过采用ChESS算子计算每个聚类对应的图像特征点,实现特征点提取.然后,提出初始角点概念,并以检测到的初始角点为起始点,对具有不同对称性的角点提出相应的检测方法.最后,以检测到的角点为初值进行角点亚像素检测.实验结果表明,检测到的角点重投影精度约为0.12个像素.提出的方法可实现棋盘图像角点自动检测,满足摄像机标定要求.     

一种棋盘格靶标不完整角点识别的修正算法

针对Hessian算子在棋盘格角点检测过程中边缘局部区域角点检测失败的问题,提出一种恢复角点识别的修正算法.首先基于坐标极值法自动识别棋盘格极点及边缘包络点,提取可以完整检测到所有包络点的2条边缘线上各自首尾及中间3个特征角点;然后利用交比不变原理求出边缘线上的消失点,并依据消失点原理求出未检测到的极点;最后根据四极点图像坐标与世界坐标的单应性关系重投影得出所有角点坐标.实验结果表明,与常用棋盘格角点检测算法相比,该算法具有较强的鲁棒性,处理过程可自动实现,在摄像机标定的应用上具有较高的实用价值.     

基于改进Harris角点算法的摄像机标定技术研究

针对传统的Harris角点提取算法对噪声敏感,执行效率低等不足,提出了一种改进的Harris角点提取方法;以Harris角点提取算法为基础,通过优化角点响应函数,消除系数K,提高角点提取算法的执行效率;采用改进的 Harris 算法对黑白棋盘表格进行内角点提取,利用张正友摄像机标定法计算USB摄像机的内外参数;实验结果证明了该方法可以有效的对黑白棋盘格内角点进行提取,准确高效,提高了摄像机的标定精度和准确性,是一种可行的摄像机标定方法。     

基于多方向对称和匀质约束的SUSAN棋盘格角点检测算法

棋盘格角点检测是摄像机标定过程中的一个重要步骤。针对原有SUSAN角点检测算法在棋盘格角点检测应用中边缘点与角点难以区分的问题,通过对棋盘格角点特性的分析和实验,提出了一种基于多方向对称和匀质约束的SUSAN棋盘格角点检测算法。通过在原有SUSAN算法的基础上添加对称约束算子和匀质约束算子滤除SU- SAN检测结果中的边缘点,最终确定角点的坐标。实验结果表明,该算法在角点检测的有效性和精度方面优于已有算法。     

自适应的Harris棋盘格角点检测算法

针时摄像机标定中的棋盘格角点检测问题,分析了Harris角点检测算法局限性,综合考虑了多个参数对角点提取的影响,根据每个参数影响程度的不同采取分步调整策略,实现棋盘格角点的自适应检测算法.该算法计算量小、易于实现,提高了运算效率,有效避免了角点聚簇和角点分布不均的角点空缺现象的产生.实验结果表明,该算法对摄像机各种畸变情况下的棋盘格角点可准确提取.     

采用棋盘格靶标的全自动相机标定方法

为了去除相机标定过程中的人为干预,提出了一种采用改进的棋盘格靶标的全自动相机标定方法。识别出每幅标定图像中的四个标志圆,利用四个标志圆圆心的图像坐标和物理坐标计算二维射影变换矩阵。依据该射影变换矩阵计算出棋盘格角点的初始图像位置,接着提取亚像素精度的角点位置。迭代求解需要标定的相机参数。由实验可知,该全自动相机标定方法的棋盘格角点识别能力和相机标定精度,与Bouguet的相机标定工具箱相当,且可以显著地减少标定时间和工作量。利用20幅分辨率为640×480的靶标图像标定相机仅需16 s。     

基于Halcon标定靶的双目相机高精度标定方法

基于张氏标定原理的传统双目相机标定方法在单一背景下标定精度高,但在复杂背景下不能精确提取特征点坐标,导致相机标定精度受到影响.文中提出一种基于Halcon标定靶感兴趣区域约束的透视变换方法,可以从复杂背景中精确提取特征点坐标,提高双目相机标定精度.首先通过改进的角点提取方法从背景中提取Halcon标定靶上的感兴趣区域,...     

基于Harris算法的黑白棋盘格角点检测

简述Harris角点检测算法的原理、算法的执行流程和黑白棋盘格亚像素坐标求取的基本方法.通过实验检测出了黑白棋盘格中的角点,得到了角点坐标,并运用二次曲面拟合法得到了角点的亚像素坐标.简要分析Harris角点检测算法中各参数对检测结果的影响,给出了参数设置的一些心得.实验结果可以用于摄像机的标定.     

Symmetry Detection Using Feature Lines

In this paper, we describe a new algorithm for detecting structural redundancy in geometric data sets. Our algorithm computes rigid symmetries, i.e., subsets of a surface model that reoccur several times within the model differing only by translation, rotation or mirroring. Our algorithm is based on matching locally coherent constellations of feature lines on the object surfaces. In comparison to previous work, the new algorithm is able to detect a large number of symmetric parts without restrictions to regular patterns or nested hierarchies. In addition, working on relevant features only leads to a strong reduction in memory and processing costs such that very large data sets can be handled. We apply the algorithm to a number of real world 3D scanner data sets, demonstrating high recognition rates for general patterns of symmetry.     

Coding and generating complex periodic patterns

This paper presents a method of coding and generating infinite classes of comlex periodic geometric pattern found in various cultures. The method uses a pattern code for the known symmetry groups and is based on the infinite classes of subdivisions of the fundamental regions of each symmetry group. The subdivisions provide a skeletal grid from which selected point, line or polygonal sequences can be extracted to produce an infinite variety of geometric patternswithin a symmetry group. Surprisingly, many complex architectural patterns from different periods and cultures, notably those from historic Islamic buildings, and Chinese window lattices, can be systematically obtained by this method. The pattern-codes provide a basis for generation of simple and complex geometric patterns, and their transformation from one to another. Some directions for systematic transformations based on meta-structural principles are outlined.     

Excluding Symmetries in Constraint-Based Search

We introduce a new method, called symmetry excluding search (SES), for excluding symmetries in constraint based search. To our knowledge, it is the first declarative method that can be applied to arbitrary symmetries. The SES-method is based on the notion of symmetric constraints, which are used in our modification of a general constraint based search algorithm. The method does not influence the search strategy. Furthermore, it can be used with either the full set of symmetries, or a subset of all symmetries.We proof correctness, completeness and symmetry exclusion properties of our method. Then, we show how to apply the SES-method in the special case of geometric symmetries (rotations and reflections) and permutation symmetries. Furthermore, we give results from practical applications.     

Fast global and partial reflective symmetry analyses using boundary surfaces of mechanical components

Axisymmetry and planar reflective symmetry properties of mechanical components can be used throughout a product development process to restructure the modeling process of a component, simplify the computation of tool path trajectories, assembly trajectories, etc. To this end, the restructured geometric model of such components must be at least as accurate as the manufacturing processes used to produce them, likewise their symmetry properties must be extracted with the same level of accuracy to preserve the accuracy of their geometric model. The proposed symmetry analysis is performed on a B-Rep CAD model through a divide-and-conquer approach over the boundary of a component with faces as atomic entities. As a result, it is possible to identify rapidly all global symmetry planes and axisymmetry as well as local symmetries. Also, the corresponding algorithm is fast enough to be inserted in CAD/CAM operators as part of interactive modeling processes, it performs at the same level of tolerance than geometric modelers and it is independent of the face and edge parameterizations.     

一种基于中心对称性的聚类算法

提出了一种新型的聚类算法。这个新型的聚类算法是基于中心对称的概念之上的。使用这种基于中心对称性的聚类算法，在一个指定的数据集中的超球面形状的聚类能够被侦测出来。在对超球面性状的目标的侦测方面，这种聚类算法大大优于传统的算法。这个算法可以用于数据聚类和人脸识别方面，实验结果也证明了该算法的效果。     

基于几何约束投票的图像特征匹配

为了改进单纯依靠相似度的图像特征匹配效果,提出了一种特征匹配方法,采用投票的方法在特征匹配的过程中引入几何约束。首先根据特征的描述向量进行初步的快速匹配,得到候选特征对以后通过投票的手段在特征之间相似度矩阵中引入一种鲁棒的几何约束,并通过自适应阈值过滤的方法获得图像特征匹配。在实验中验证了该方法对图像特征匹配正确与否具有较强的区分能力,在与已有技术接近的计算时间内得到了更高的匹配准确度。基于几何约束投票的图像特征匹配比单纯依靠相似度具有更好的匹配效果。     

Perceptually guided polygon reduction

The properties of the human visual system are taken into account, along with the geometric aspects of an object, in a new surface remeshing algorithm and a new mesh simplification algorithm. Both algorithms have a preprocessing step and are followed by the remeshing or mesh simplification steps. The preprocessing step computes an importance map that indicates the visual masking potential of the visual patterns on the surface. The importance map is then used to guide the remeshing or mesh simplification algorithms. Two different methods are proposed for computing an importance map that indicates the masking potential of the visual patterns on the surface. The first one is based on the Sarnoff visual discrimination metric, and the second one is inspired by the visual masking tool available in the current JPEG2000 standard. Given an importance map, the surface remeshing algorithm automatically distributes few samples to surface regions with strong visual masking properties due to surface texturing, lighting variations, bump mapping, surface reflectance and inter-reflections. Similarly, the mesh simplification algorithm simplifies more aggressively where the light field of an object can hide more geometric artifacts.     

一种新的棋盘格图像角点检测算法

鉴于棋盘格图案在摄像机标定中有着广泛的应用,因此为了对边缘模糊的棋盘格图像进行在线标定,针对目前棋盘格图像角点检测算法的局限性,提出了一种可以自动实现的棋盘格图像角点检测算法,该算法是利用周围图案对称和灰度值对比明显的独特性质,设计了由对称算子S和方差算子V组合而成的角点检测算子--对称方差算子（symmetry and variance）,简称SV算子.该算子不仅构思巧妙、易于实现、计算量小,而且对棋盘格图像的旋转变换和亮度变换具有鲁棒性和抗噪能力强的优点.实验结果表明,该算子对边缘模糊有良好的适应性,适用于摄像机的在线标定.     

基于加权对称图像的二维FDA人脸识别算法

提出一种融合加权对称图像的二维FDA人脸识别算法。将人脸图像分解为奇偶对称脸,并利用加权因子将奇偶对称脸重构新的人脸样本,通过二维FDA算法求解新样本图像的最优特征子空间进行人脸分类。有效融合二维FDA算法的优点,并利用人脸对称性的特征,同时进一步分析加权因子对人脸识别效果的影响,通过选取最优加权因子最大地提高识别率。在人脸图像库ORL中进行的实验结果表明,该算法有效并能获得较高的识别率。