一种基于角点检测的图像密集匹配算法

提出了一种鲁棒的图像自动立体匹配算法.利用Sobel算子对图像中的像素点进行检测,若是边缘点,则使用最小同值分割吸收核方法判断该点是否为角点.在两幅待匹配的图像间计算角点的梯度大小、梯度方向及灰度等的相似度,去除无法对应的角点,建立起待匹配图像中角点的对应关系,并计算基础矩阵.对基础矩阵进行迭代,去除误配点,计算出较精确的基础矩阵.由对极几何约束,采用动态规划方法,寻找左右两幅图像在对应极线上的所有像素点之间的对应,从而建立起两幅图像间像素点的密集匹配对应关系.试验结果表明,算法效果满意.     

三视校正的理论及鲁棒性算法

主要讨论两方面的工作.首先,对三视校正问题进行理论分析,给出了校正后图像的基本矩阵与其约束条件之间的关系,讨论了三视校正过程中的6个自由参数的几何含义.这些结果为处理校正过程中带来的图像射影畸变提供了理论根据.其次,在RANSAC(random sampling consensus)框架下,提出了一种鲁棒的三视校正算法.与传统的校正算法不同,该算法不再只依赖于基本矩阵,而是直接利用了原始匹配点的信息.这种基于点的方法有两个优点:一方面,由于噪声的干扰,基本矩阵往往估计得不够准确;另一方面,由于基本矩阵的评价准则与校正结果的评价准则不同,即使从好的基本矩阵出发,也未必能获得好的校正结果.大量的模拟和真实图像实验表明,该算法具有很强的抗噪声及抗错误匹配的能力,能够获得令人满意的校正效果.     

基于三视图约束的基础矩阵估计

考虑到只依赖对极几何关系的匹配点余差并不能完全区分匹配点的正确与否,从而影响内点集选取的情况,提出基于三视图约束的基础矩阵估计算法。首先,使用传统随机抽样一致性（RANSAC）算法计算三视图的任意两对相邻图像间的基础矩阵,确定三视图中共有的匹配点对,并计算估计基础矩阵时非共用图像上的匹配点在共用图像上的极线;然后,计算两条极线的交点与共用图像上对应匹配点间的距离,以距离值的大小作为内点判断的依据,得到新的内点集。在新内点集的基础上,采用M估计算法重新计算基础矩阵。实验结果表明：该方法可以同时降低噪声和错误匹配对基础矩阵精确计算的影响,精度优于传统鲁棒性算法,使点到极线的距离限制在0.3个像素左右,而且计算结果具有稳定性,可以被广泛地应用到基于图像序列的三维重建和摄影测量等领域中。     

基于遗传算法不同策略下的基础矩阵估计方法

在未定标系统中，对极几何约束给出了图像间的全部信息，成为解决许多视觉问题的关键环节，提出了一种基于遗传算法不同策略下的基础矩阵估计方法，它利用每个基因代表一个匹配点，每条染色体作为基础矩阵计算时的最小子集，并根据染色体长度决定采用何种策略估计基础矩阵，此方法在很大程度上减小了出格点对估计过程的影响，能够较好地汇聚到全局最优解，模拟数据和真实图像的实验结果都表明，所给出的方法能够有效地检测和删除错定位和误匹配点，提高了基础矩阵估计的鲁棒性和精度。     

视觉基础矩阵估计方法的性能比较与分析

基础矩阵描述了单个场景的2幅视图之间的对应关系,在计算机视觉领域中扮演着极其重要的角色.首先讨论了三维视觉系统中的极线几何理论,随后论述了几类基础矩阵的估计方法,并对这些方法作了详细的比较和评价,最后实现了各种算法且使用仿真数据以及真实图像数据对各自的性能作了分析和总结.比较结果说明:1)如果图像特征点定位精确并且匹配无误,则线性方法的结果相当好;2)迭代算法可以解决定位的高斯噪声,但是当数据存在错误匹配时,性能很差;3)鲁棒算法能够同时解决定位误差和误匹配两类问题.此外,模拟实验和真实图像实验的结果表明,基于特征分析的正交回归最小二乘法的计算结果优于经典的线性最小二乘法.     

基于SIFT特征和对极几何约束的图像配准算法

基于SIFT（尺度不变特征变换）特征匹配思想,提出了一种应用对极几何约束的图像特征配准算法。首先对图像提取SIFT特征点,然后通过欧氏距离估算对SIFT特征描述子进行初始匹配得到预匹配点集;采用基于单应矩阵的抽样算法计算初始基础矩阵,通过RANSAC算法计算精确的基础矩阵和匹配点集,进而实现图像配准。实验表明,该算法可以获得更准确的匹配点,得到精度较高的图像配准效果。     

基于Sampson误差计算单应的立体匹配

立体匹配是立体视觉重要的研究内容,文章在摄像机无标定的前提下采用Sampson误差估计立体图像对的平面单应,由单应计算基础矩阵,进行图像特征点匹配。先用Forstner算子提取角点,按照灰度差相似性准则进行初始匹配,然后在Sampson误差模型下求解代价函数的最优解,解决单应矩阵元素方程组超定问题。在平面单应及对极几何约束下进行图像特征点匹配,可获得射影意义下的象点重构。实验结果表明该方法能准确快速地匹配特征点。     

基于SIFT特征点的图像拼接技术研究

给出一种基于特征点的图像拼接方法,该方法采用对于尺度具有鲁棒性的SIFT算法进行特征点的提取与匹配,计算出特征匹配点后,使用RANSAC算法剔除误配,并计算出两幅图像之间的坐标变换关系矩阵H.最终使用加权平滑算法完成了图像的无缝拼接.实验证明,该算法有效提高了图像拼接的效率和准确性.     

基于运动平滑约束项的快速误匹配剔除算法

针对图像拼接时用随机抽样一致性（RANSAC）算法迭代计算过程中计算量大、匹配正确率低的问题,提出了一种基于运动平滑约束项的误匹配剔除算法。首先采用快速旋转不变特征（ORB）算法提取特征点,基于汉明距离实现特征点初匹配;其次,基于运动平滑约束项统计邻域支持估计量实现误匹配粗剔除;然后,进一步采用空间几何约束关系实现误匹配精剔除;最后,利用分组排序采样求解模型参数,采用加权平均实现图像融合。实验结果表明,该算法的误匹配剔除率相比缩小抽样点总量算法提升了75.6%,相比自适应阈值算法提升了24%,此方法能有效剔除误匹配,实现图像精确拼接。     

基于仿射点对应的分层重构

提出了一种基于仿射点对应的分层重构方法,所谓仿射点对应是指相差一个仿射变换的两个空间点集的图像对应．该方法主要分为以下三个步骤：首先,从点对应计算准仿射重构;然后,由仿射点对应的准仿射重构建立一个三维射影变换,并利用这个射影变换的特征向量来确定无穷远平面,从而得到仿射重构;最后,从仿射重构所获得的无穷远平面单应矩阵标定摄像机内参数,进而得到度量重构．在上述三个步骤中,第二个步骤是最关键的,即如何确定对应于无穷远平面的特征向量,这也是该文的新思想和主要贡献所在．仿真和真实图像实验均表明,该文的方法是有效的,并且有很好的鲁棒性．     

非标定立体图像对鲁棒性校正新算法

传统的非标定图像校正都是依赖于基础矩阵的精度。而该方法根据图像校正基本准则，提出一种无需高精度基础矩阵的鲁棒性图像对校正算法，基础矩阵只是提供迭代优化的初始数值，并且与需要基础矩阵的方法相比较，结果表明该方法有效地消除垂直视差，能够较好地处理射影畸变问题。     

Estimating the fundamental matrix based on least absolute deviation

The epipolar geometry is the intrinsic projective geometry between two views, and the algebraic representation of it is the fundamental matrix. Estimating the fundamental matrix requires solving an over-determined equation. Many classical approaches assume that the error values of the over-determined equation obey a Gaussian distribution. However, the performances of these approaches may decrease significantly when the noise is large and heterogeneous. This paper proposes a novel technique for estimating the fundamental matrix based on least absolute deviation (LAD), which is also known as the L₁ method. Then a linear iterative algorithm is presented. The experimental results on some indoor and outdoor scenes show that the proposed algorithm yields the accurate and robust estimates of the fundamental matrix when the noise is non-Gaussian.     

Projective reconstruction and invariants from multiple images

This correspondence investigates projective reconstruction of geometric configurations seen in two or more perspective views, and the computation of projective invariants of these configurations from their images. A basic tool in this investigation is the fundamental matrix that describes the epipolar correspondence between image pairs. It is proven that once the epipolar geometry is known, the configurations of many geometric structures (for instance sets of points or lines) are determined up to a collineation of projective 3-space 𝒫³ by their projection in two independent images. This theorem is the key to a method for the computation of invariants of the geometry. Invariants of six points in 𝒫³ and of four lines in 𝒫³ are defined and discussed. An example with real images shows that they are effective in distinguishing different geometrical configurations. Since the fundamental matrix is a basic tool in the computation of these invariants, new methods of computing the fundamental matrix from seven-point correspondences in two images or six-point correspondences in three images are given     

Theory and Practice of Projective Rectification

This paper gives a new method for image rectification, the process of resampling pairs of stereo images taken from widely differing viewpoints in order to produce a pair of matched epipolar projections. These are projections in which the epipolar lines run parallel with the x-axis and consequently, disparities between the images are in the x-direction only. The method is based on an examination of the fundamental matrix of Longuet-Higgins which describes the epipolar geometry of the image pair. The approach taken is consistent with that advocated by Faugeras (1992) of avoiding camera calibration. The paper uses methods of projective geometry to determine a pair of 2D projective transformations to be applied to the two images in order to match the epipolar lines. The advantages include the simplicity of the 2D projective transformation which allows very fast resampling as well as subsequent simplification in the identification of matched points and scene reconstruction.     

From lines to epipoles through planes in two views

This paper addresses the computation of the fundamental matrix between two views, when camera motion and 3D structure are unknown, but planar surfaces can be assumed. We use line features which are automatically matched in two steps. Firstly, with image based parameters, a set of matches are obtained to secondly compute homographies, which allows to reject wrong ones, and to grow good matches in a final stage. The inclusion of projective transformations gives much better results to match features with short computing overload. When two or more planes are observed, different homographies can be computed, segmenting simultaneously the corresponding planar surfaces. These can be used to obtain the fundamental matrix, which gives constraints for the whole scene. The results show that the global process is robust enough, turning out stable and useful to obtain matches and epipolar geometry from lines in man made environments.     

基于标准脸型的人脸3D姿态估计方法

针对二维人像的三维姿态估计,结合标准脸型、人脸的统计知识和射影几何提出一种方法：多次选取标准脸型眼角点、嘴角点、鼻下点等特征点得到特征三角形并确定人脸三维模型;其次计算人脸三维模型参数;基于透视投影模型计算旋转矩阵;基于最小二乘法计算三维姿态。提取的特征点易于标定,无需任何辅助设备,具有一定的准确性。     

The quasi-perspective model: Geometric properties and 3D reconstruction

The paper investigates geometric properties of quasi-perspective projection model in one and two-view geometry. The main results are as follows. (i) Quasi-perspective projection matrix has nine degrees of freedom (DOF), and the parallelism along X and Y directions in world system are preserved in images. (ii) Quasi-fundamental matrix can be simplified to a special form with only six DOFs. The fundamental matrix is invariant to any non-singular projective transformation. (iii) Plane induced homography under quasi-perspective model can be simplified to a special form defined by six DOFs. The quasi-homography may be recovered from two pairs of corresponding points with known fundamental matrix. (iv) Any two reconstructions in quasi-perspective space are defined up to a non-singular quasi-perspective transformation. The results are validated experimentally on both synthetic and real images.