Three-dimensional data acquisition by trinocular vision

For recognition of three-dimensional (3D) shapes and measurement of 3D positions of objects it is important for a vision system to be able to measure the 3D data of dense points in the environment. One approach is to measure the distance on the basis of the triangulation principle from the disparity of two images. However, this binocular vision method has difficulty in finding a correspondence of features between two images. This correspondence problem can be solved geometrically by adding another camera, i.e. by trinocular vision. This paper presents the principles and implementation details of trinocular vision. On the basis of the proposed method, we carried out several experiments, from which we found that many correct correspondences could be established, even for images of a complex scene, by only the geometrical constraint of trinocular vision. However, when there are dense points in the image, multiple candidate points are found and a unique correspondence cannot be established. Two approaches to solve this problem are discussed in this paper.     

一种基于基本矩阵的相机畸变差自动校正方法

在计算机视觉的应用领域中,为了提高图像量测和3维重建的精度,必须对相机的畸变误差进行修正。为此提出了一种基于基本矩阵的相机径向畸变的自动校正方法,该方法不需要预先获得场景的结构信息和相机的内部参数,仅利用两张影像同名点集之间的内在几何关系,即可求取相机的径向畸变系数,进而可对这两幅图像的畸变误差进行自动校正。试验结果表明,该方法是一种有效的畸变图像校正算法,能够获得到满意的校正结果。     

Appearance-based virtual view generation from multicamera videos captured in the 3-D room

We present an appearance-based virtual view generation method that allows viewers to fly through a real dynamic scene. The scene is captured by multiple synchronized cameras. Arbitrary views are generated by interpolating two original camera-views near the given viewpoint. The quality of the generated synthetic view is determined by the precision, consistency and density of correspondences between the two images. All or most of previous work that uses interpolation extracts the correspondences from these two images. However, not only is it difficult to do so reliably (the task requires a good stereo algorithm), but also the two images alone sometimes do not have enough information, due to problems such as occlusion. Instead, we take advantage of the fact that we have many views, from which we can extract much more reliable and comprehensive 3D geometry of the scene as a 3D model. Dense and precise correspondences between the two images, to be used for interpolation, are obtained using this constructed 3D model.     

Stereo-motion with stereo and motion in complement

This paper presents a new approach of combining stereo vision and dynamic vision with the objective of retaining their advantages and removing their disadvantages. It is shown that, by assuming affine cameras, the stereo correspondences and motion correspondences, if organized in a particular way in a matrix, can be decomposed into: the 3D structure of the scene, the camera parameters, the motion parameters, and the stereo geometry. With this, the approach can infer stereo correspondences from motion correspondences, requiring only a time linear with respect to the size of the available image data. The approach offers the advantages of simpler correspondence, as in dynamic vision, and accurate reconstruction, as in stereo vision, even with short image sequences     

Robust active stereo vision using Kullback-Leibler divergence

Active stereo vision is a method of 3D surface scanning involving the projecting and capturing of a series of light patterns where depth is derived from correspondences between the observed and projected patterns. In contrast, passive stereo vision reveals depth through correspondences between textured images from two or more cameras. By employing a projector, active stereo vision systems find correspondences between two or more cameras, without ambiguity, independent of object texture. In this paper, we present a hybrid 3D reconstruction framework that supplements projected pattern correspondence matching with texture information. The proposed scheme consists of using projected pattern data to derive initial correspondences across cameras and then using texture data to eliminate ambiguities. Pattern modulation data are then used to estimate error models from which Kullback-Leibler divergence refinement is applied to reduce misregistration errors. Using only a small number of patterns, the presented approach reduces measurement errors versus traditional structured light and phase matching methodologies while being insensitive to gamma distortion, projector flickering, and secondary reflections. Experimental results demonstrate these advantages in terms of enhanced 3D reconstruction performance in the presence of noise, deterministic distortions, and conditions of texture and depth contrast.     

基于图像序列的交互式快速建模系统

给出了一个基于图像序列的交互式三维建模系统.通过输入一段未标定的图像或视频序列,系统能够自动地恢复出摄像机参数;然后用户只需要在少量几帧图像上简单勾画出物体的形态结构,系统就能自动解析出多帧之间用户交互的对应关系,从而迅速、逼真地重建出场景的三维模型.该系统提供了点与线段的重建、直线与平面的重建、曲线与曲面的重建等功能,能够满足对现实世界中的复杂场景的快速高精度的重建要求.几组真实拍摄的图像序列的建模实验表明:该系统高效、实用.能够很好地满足实际建模需求.     

基于条纹边界编码光系统的三维分层重构技术

给出了一种基于条纹编码光的三维分层重建技术;为了提高结构编码光的抗干扰性,采用一种基于时空多帧编码的投射条纹图像,通过整合分析多帧图像不同条纹间的跳变关系得到被测物象的光栅条纹编码从而有效提高结构光的抗干扰性;采用一种新的基于组合数学的编码算法以利于条纹编码自动化;在既有文献基础上,提出新的基于条纹边界编码光的三维分层重构数学模型。实验结果证明,在3D重建方面,该技术具有较强的鲁棒性。     

一种基于主动视觉的三维结构恢复和直接欧氏重建算法

利用三正交平移运动, 提出了一种三维结构恢复和直接欧氏重建新算法. 算法仅需利用主动视觉平台控制相机作一组三正交平移运动, 然后通过图像对应点和平移运动的距离就可以恢复平面结构信息和进行欧氏重建. 并且无需假定相机畸变因子为零. 算法计算过程中无需求解相机的内参数, 也无需进行分层重构, 它是一种直接的欧氏重建算法, 避免了传统算法中的相机标定、仿射重建等两大难题, 并且计算过程完全线性化, 简单实用. 最后用模拟实验和真实图像实验对算法进行验证, 实验结果表明了算法的有效性和准确性.     

Design of an improved framework for stratified 3D reconstruction from a pair of images,with reduced ambiguity

Reconstruction of three dimensional (3D) object structure from multiple images is a fundamental problem in computational vision. Many applications in computer vision require the use of structure information of 3D objects. The objective of this work is to develop a stable method of 3D reconstruction of an object, which works without the availability of camera parameters, once the plane at infinity is obtained using the approximate scene information. First, a framework has been designed based on a modification of the auto-calibration procedure for 3D structure computation using singular value decomposition. In the second part of the work, ambiguities present at the various stages of 3D reconstruction have been analysed. Error norms have been proposed, and studied to quantify the ambiguity in the reconstruction process. We attempt to analyse the effect of pose difference between camera views and focal length parameters on the reconstruction process, using experimentation with simulated and real-world data.     

Two-view correspondence learning via complex information extraction

Establishing reliable correspondences plays a vital role in many feature-matching based computer vision tasks. Given putative correspondences of feature points in two images, in this paper, we propose a novel network for inferring the probabilities of correspondences being inliers or outliers and regressing the relative pose encoded by the essential matrix. Previous research proposed an end-to-end permutation-equivariant classification network based on multi-layer perceptrons and context normalization. However, the context normalization treats each correspondence equally and ignore the extraction of channel information, as a result the representation capability of potential inliers can be reduced. To solve this problem, we apply attention mechanism in our network to capture complex information of the feature maps. Specifically, we introduce two types of attention blocks. We adopt the spatial attention block to capture complex spatial contextual information, and the rich channel information can be obtained by utilizing the channel attention block. To obtain richer contextual information and feature maps with stronger representative capacity, We combine these attention blocks with the PointCN block to form a new network with strong representative ability. Experimental results on several benchmark datasets show that the performance on outlier removal and camera pose estimation is significantly improved over the state-of-the-arts.

     

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．     

基于弱监督学习的三维人脸形状与纹理重建

三维人脸相较于二维人脸包含了更多特征信息, 可应用于如人脸识别、影视娱乐、医疗美容等更多实际应用场景, 因此三维人脸重建技术一直是计算机视觉领域的研究热点. 由于真实三维人脸数据较难获取, 很多基于深度学习的重建算法首先利用传统重建方法为大量二维人脸图像构建三维标签, 作为训练数据, 这些数据可能并不精准, 从而导致算法的重建精度受到影响. 为此, 本文提出一种基于multi-level损失函数的弱监督学习模型, 结合传统三维人脸形变模型3DMM与深度学习方法, 直接从大量无三维标签的二维人脸图像中学习三维人脸特征信息, 从而实现基于单张二维人脸图像的三维人脸重建算法. 此外, 为解决二维人脸图像中常存在遮挡或大姿态情况而影响人脸纹理重建的问题, 本文使用基于CelebAMask-HQ数据集的人脸解析分割算法对图像进行预处理去除遮挡区域. 实验结果表明, 基于本文方法的三维人脸重建质量与重建精度均实现了一定的提升.     

由平行平面的投影确定无穷远平面的单应矩阵

在三维计算机视觉中,无穷远平面的单应矩阵扮演了极其重要的角色,可使众多视觉问题的求解得到简化.主要讨论如何利用平行平面的投影来求解两个视点间的无穷远平面的单应矩阵,用代数方法构造性地证明了下述结论:(1) 如果场景中含有一组平行平面,则可以通过求解一个一元4次方程来确定两个视点间的无穷远平面对应的单应矩阵;(2) 如果场景中含有两组平行平面,则可以线性地确定两个视点间的无穷远平面对应的单应矩阵.并对上述结果给出了相应的几何解释和具体算法.所给出的结果在三维计算机视觉,特别是摄像机自标定中具有一定的理论意义和应用价值.     

Binocular matching constraints from motion

Structure from motion and structure from stereo are two vision cues for achieving 3D reconstruction. The two cues have complementary strengths; while 3D reconstruction is accurate but correspondence establishment is difficult in the stereo cue, the reverse is true in the motion cue. This paper addresses how to combine the two cues when a stereo pair of cameras are available to capture image data for 3D reconstruction. The work is distinct in that, in contrast with the previous ones, it is not to exploit the redundancy in the image data for boosting the reconstruction accuracy, but to make the two vision cues complementary, preserving their strengths and avoiding their weaknesses. A mechanism is introduced that allows dense motion correspondences in the two separate image streams be transferred to dense binocular correspondences across the image streams, so that 3D can be reconstructed from the latter and accurate reconstruction is possible even with short motions of the stereo rig. Both the stereo correspondences and the motion of the stereo rig are assumed to be unknown in this work. Experiments involving real image data are presented to indicate the feasibility and robustness of the approach.     

3D Video Billboard Clouds

3D video billboard clouds reconstruct and represent a dynamic three-dimensional scene using displacement-mapped billboards. They consist of geometric proxy planes augmented with detailed displacement maps and combine the generality of geometry-based 3D video with the regularization properties of image-based 3D video. 3D video billboards are an image-based representation placed in the disparity space of the acquisition cameras and thus provide a regular sampling of the scene with a uniform error model. We propose a general geometry filtering framework which generates time-coherent models and removes reconstruction and quantization noise as well as calibration errors. This replaces the complex and time-consuming sub-pixel matching process in stereo reconstruction with a bilateral filter. Rendering is performed using a GPU-accelerated algorithm which generates consistent view-dependent geometry and textures for each individual frame. In addition, we present a semi-automatic approach for modeling dynamic three-dimensional scenes with a set of multiple 3D video billboards clouds.     

Real-time exploration and photorealistic reconstruction of large natural environments

This paper presents a hybrid (geometry- and image-based) framework suitable for providing photorealistic walkthroughs of large, complex outdoor scenes, based only on a small set of real images from the scene. To this end, a novel data representation of a 3D scene is proposed, which is called morphable 3D panoramas. Motion is assumed to be taking place along a predefined path of the 3D environment and the input to the system is a sparse set of stereoscopic views at certain positions (key positions) along that path (one view per position). An approximate local 3D model is constructed from each view, capable of capturing the photometric and geometric properties of the scene only locally. Then, during the rendering process, a continuous morphing (both photometric as well as geometric) takes place between successive local 3D models, using what we call a ‘morphable 3D model’. For the estimation of the photometric morphing, a robust algorithm capable of extracting a dense field of 2D correspondences between wide-baseline images is used, whereas, for the geometric morphing, a novel method of computing 3D correspondences between local models is proposed. In this way, a physically valid morphing is always produced, which is thus kept transparent from the user. Moreover, a highly optimized rendering path is used during morphing. Thanks to the use of appropriate pixel and vertex shaders, this rendering path can be run fully in 3D graphics hardware and thus allows for high frame rates. Our system can be extended to handle multiple stereoscopic views (and therefore multiple local models) per key position of the path (related by a camera rotation). In this case, one local 3D panorama (per key position) is constructed, comprising all local 3D models therein, and so a ‘morphable 3D panorama’ is now used during the rendering process. For handling the geometric consistency of each 3D panorama, a technique which is based on solving a partial differential equation is adopted. The effectiveness of our framework is demonstrated by using it for the 3D visual reconstruction of the Samaria Gorge in Crete.     

Three-Dimensional Reconstruction of Points and Lines with Unknown Correspondence across Images

Three-dimensional reconstruction from a set of images is an important and difficult problem in computer vision. In this paper, we address the problem of determining image feature correspondences while simultaneously reconstructing the corresponding 3D features, given the camera poses of disparate monocular views. First, two new affinity measures are presented that capture the degree to which candidate features from different images consistently represent the projection of the same 3D point or 3D line. An affinity measure for point features in two different views is defined with respect to their distance from a hypothetical projected 3D pseudo-intersection point. Similarly, an affinity measure for 2D image line segments across three views is defined with respect to a 3D pseudo-intersection line. These affinity measures provide a foundation for determining unknown correspondences using weighted bipartite graphs representing candidate point and line matches across different images. As a result of this graph representation, a standard graph-theoretic algorithm can provide an optimal, simultaneous matching and triangulation of points across two views, and lines across three views. Experimental results on synthetic and real data demonstrate the effectiveness of the approach.An erratum to this article can be found at     

医学3D计算机视觉：研究进展和挑战

医学影像的诊断是许多临床决策的基础,而医学影像的智能分析是医疗人工智能的重要组成部分。与此同时,随着越来越多3D空间传感器的兴起和普及,3D计算机视觉正变得越发重要。本文关注医学影像分析和3D计算机的交叉领域,即医学3D计算机视觉或医学3D视觉。本文将医学3D计算机视觉系统划分为任务、数据和表征3个层面,并结合最新文献呈现这3个层面的研究进展。在任务层面,介绍医学3D计算机视觉中的分类、分割、检测、配准和成像重建,以及这些任务在临床诊断和医学影像分析中的作用和特点。在数据层面,简要介绍了医学3D数据中最重要的数据模态：包括计算机断层成像（computed tomography,CT）、磁共振成像（magnetic resonance imaging,MRI）、正电子放射断层成像（positron emission tomography,PET）等,以及一些新兴研究提出的其他数据格式。在此基础上,整理了医学3D计算机视觉中重要的研究数据集,并标注其数据模态和主要视觉任务。在表征层面,介绍并讨论了2D网络、3D网络和混合网络在医学3D数据的表征学习上的优缺点。此外,针对医学影像中普遍存在的小数据问题,重点讨论了医学3D数据表征学习中的预训练问题。最后,总结了目前医学3D计算机视觉的研究现状,并指出目前尚待解决的研究挑战、问题和方向。     

Self-Calibration Under the Cayley Framework

The Cayley framework here is meant to tackle the vision problems under the infinite Cayley transformation (ICT), its main advantage lies in its numerical stability. In this work, the stratified self-calibration under the Cayley framework is investigated. It is well known that the main difficulty of the stratified self-calibration in multiple view geometry is to upgrade a projective reconstruction to an affine one, in other words, to estimate the unknown 3-vector of the plane at infinity, called the normal vector. To our knowledge, without any prior knowledge about the scene or the camera motion, the only available constraint on a moving camera with constant intrinsic parameters is the well-known Modulus Constraint in the literature. Do other kinds of constraints exist? If yes, what they are? How could they be used? In this work, such questions will be systematically investigated under the Cayley framework. Our key contributions include: 1. The original projective expression of the ICT is simplified and a new projective expression is derived to make the upgrade easier from a projective reconstruction to a metric reconstruction. 2. The constraints on the normal vector are systematically investigated. For two views, two constraints on the normal vector are derived; one of them is the well-known modulus constraint, while the other is a new inequality constraint. There are only these two constraints for two views. For three views, besides the constraints for two views, two groups of new constraints are derived and each of them contains three constraints. In other words, there are 12 constraints in total for three views. 3. Based on our projective expression and these constraints, a stratified Cayley algorithm and a total Cayley algorithm are proposed for the metric reconstruction from images. It is experimentally shown that they both improve significantly the numerical stability of the classical algorithms. Compared with the global optimal algorithm under the infinite homography framework, the Cayley algorithms have comparable calibration accuracy, but substantially reduce the computational load.     

宽基线视差图像的拼接算法

针对两幅视差图像的拼接问题,提出了一种新算法,即利用Hessian仿射不变检测算子检测出特征区域,利用SIFT特征描述算子提取特征区域特征矢量,根据特征矢量的欧几里德距离来建立图像间的稀疏对应关系;由这些对应点稀疏地确定场景中的一些点,以这些点为顶点建立场景的三角面片近似,再据此将重叠区域重投影生成推扫式成像的中间部分图像。将中间部分推扫式成像图像和原左图像的左半部分以及原右图像的右半部分一起拼接生成大图像。利用实际图像进行的拼接实验表明该算法是一个有效的视差图像拼接算法。