Multi-view structure-from-motion for hybrid camera scenarios

We describe a pipeline for structure-from-motion (SfM) with mixed camera types, namely omnidirectional and perspective cameras. For the steps of this pipeline, we propose new approaches or adapt the existing perspective camera methods to make the pipeline effective and automatic. We model our cameras of different types with the sphere camera model. To match feature points, we describe a preprocessing algorithm which significantly increases scale invariant feature transform (SIFT) matching performance for hybrid image pairs. With this approach, automatic point matching between omnidirectional and perspective images is achieved. We robustly estimate the hybrid fundamental matrix with the obtained point correspondences. We introduce the normalization matrices for lifted coordinates so that normalization and denormalization can be performed linearly for omnidirectional images. We evaluate the alternatives of estimating camera poses in hybrid pairs. A weighting strategy is proposed for iterative linear triangulation which improves the structure estimation accuracy. Following the addition of multiple perspective and omnidirectional images to the structure, we perform sparse bundle adjustment on the estimated structure by adapting it to use the sphere camera model. Demonstrations of the end-to-end multi-view SfM pipeline with the real images of mixed camera types are presented.     

Real-time and accurate extrinsic camera parameter estimation using feature landmark database for augmented reality

In the field of augmented reality (AR), many kinds of vision-based extrinsic camera parameter estimation methods have been proposed to achieve geometric registration between real and virtual worlds. Previously, a feature landmark-based camera parameter estimation method was proposed. This is an effective method for implementing outdoor AR applications because a feature landmark database can be automatically constructed using the structure-from-motion (SfM) technique. However, the previous method cannot work in real time because it entails a high computational cost or matching landmarks in a database with image features in an input image. In addition, the accuracy of estimated camera parameters is insufficient for applications that need to overlay CG objects at a position close to the user's viewpoint. This is because it is difficult to compensate for visual pattern change of close landmarks when only the sparse depth information obtained by the SfM is available. In this paper, we achieve fast and accurate feature landmark-based camera parameter estimation by adopting the following approaches. First, the number of matching candidates is reduced to achieve fast camera parameter estimation by tentative camera parameter estimation and by assigning priorities to landmarks. Second, image templates of landmarks are adequately compensated for by considering the local 3-D structure of a landmark using the dense depth information obtained by a laser range sensor. To demonstrate the effectiveness of the proposed method, we developed some AR applications using the proposed method.     

结合视觉惯性模组的室内三维布局鲁棒重建方法

针对使用传统单目相机的全自动三维重建方法结果精确度差和整体结构理解缺失等问题,提出一种结合视觉惯性里程计和由运动到结构的全自动室内三维布局重建系统.首先利用视觉里程计获得关键帧图像序列和对应空间位置姿态,并利用运动恢复结构算法计算精确相机位姿;然后利用多图视立体几何算法生成高质量稠密点云;最后基于曼哈顿世界假设,针对典型的现代建筑室内场景,设计一种基于规则的自底向上的布局重建方法,得到最终房间外轮廓布局.使用浙江大学CAD&CG实验室场景现场扫描数据集和人工合成的稠密点云数据集作为实验数据,在Ubuntu 16.04和PCL 1.9环境下进行实验.结果表明,文中方法对三维点云噪声容忍度高,能够有效地重建出室内场景的三维外轮廓布局.     

Experiments on estimating egomotion and structure parameters using long monocular image sequences

This paper presents a simple but robust model based approach to estimating the kinematics of a moving camera and the structure of the objects in a stationary environment using long, noisy, monocular image sequences. Both batch and recursive algorithms are presented and the problem due to occlusion is addressed. The approach is based on representing the constant translational velocity and constant angular velocity of the camera motion using nine rectilinear motion parameters, which are 3-D vectors of the position of the rotation center, linear and angular velocities. The structure parameters are 3-D coordinates of the salient feature points in the inertial coordinate system. Due to redundancies in parameterization, the total number of independent parameters to be estimated is 3M+7, whereM is the number of feature points. The image plane coordinates of these feature points in each frame are first detected and matched over the frames. These noisy image coordinates serve as the input to our algorithms. Due to the nonlinear nature of perspective projection, a nonlinear least squares method is formulated for the batch algorithm, and a conjugate gradient method is then applied to find the solution. A recursive method using an Iterated Extended Kalman Filter (IEKF) for incremental estimation of motion and structure is also presented. Since the plant model is simple in our formulation, closed form solutions for the state and covariance transition equations are easily derived. Experimental results for simulated imagery as well as several real image sequences are included.The support of the Advanced Research Projects Agency (ARPA order No. 8459), the U.S. Army Topographic Engineering Center under contract DACA 76-92-C-0009, and the Department of Electrical Engineering at the University of Maryland is gratefully acknowledged.     

Fast moving object detection with non-stationary background

The detection of moving objects under a free-moving camera is a difficult problem because the camera and object motions are mixed together and the objects are often detected into the separated components. To tackle this problem, we propose a fast moving object detection method using optical flow clustering and Delaunay triangulation as follows. First, we extract the corner feature points using Harris corner detector and compute optical flow vectors at the extracted corner feature points. Second, we cluster the optical flow vectors using K-means clustering method and reject the outlier feature points using Random Sample Consensus algorithm. Third, we classify each cluster into the camera and object motion using its scatteredness of optical flow vectors. Fourth, we compensate the camera motion using the multi-resolution block-based motion propagation method and detect the objects using the background subtraction between the previous frame and the motion compensated current frame. Finally, we merge the separately detected objects using Delaunay triangulation. The experimental results using Carnegie Mellon University database show that the proposed moving object detection method outperforms the existing other methods in terms of detection accuracy and processing time.     

Automatic feature point extraction and tracking in image sequences for arbitrary camera motion

An automatic egomotion compensation based point correspondence algorithm is presented. A basic problem in autonomous navigation and motion estimation is automatically detecting and tracking features in consecutive frames, a challenging problem when camera motion is significant. In general, feature displacements between consecutive frames can be approximately decomposed into two components: (i) displacements due to camera motion which can be approximately compensated by image rotation, scaling, and translation; (ii) displacements due to object motion and/or perspective projection. In this paper, we introduce a two-step approach: First, the motion of the camera is compensated using a computational vision based image registration algorithm. Then consecutive frames are transformed to the same coordinate system and the feature correspondence problem is solved as though tracking moving objects for a stationary camera. Methods of subpixel accuracy feature matching, tracking and error analysis are introduced. The approach results in a robust and efficient algorithm. Results on several real image sequences are presented.The support of the Advanced Research Projects Agency (ARPA Order No. 8459) and the U.S. Army Engineer Topographic Laboratories under Contract DACA 76-92-C-0009 is gratefully acknowledged.     

抗相机运动干扰的欧拉视频振动检测

相机运动的干扰是造成视频振动检测误差的重要原因。针对该问题,提出一种互抑制一致采样方法,对视频中的振动信号和相机运动信号实现有效分离,从而提高视频振动检测的可靠性。通过SURF（加速稳健特征）算法提取候选的特征点,并设计了振动与相机运动的互抑制测度,对候选的特征点进行分离,以获得相机运动的特征点。根据相机运动特征点对视频图像进行配准,以获得去除相机运动干扰的视频序列。对稳定相机的视频序列,采用欧拉视频振动检测方法获得振动频率。自行采集了不同相机运动下的视频,并针对互抑制测度的参数进行估计。通过对测试集数据进行验证,得到的振动频率准确率优于现有的非接触振动检测方法。     

An adaptive observer framework for accurate feature depth estimation using an uncalibrated monocular camera

This paper presents a novel solution to the problem of depth estimation using a monocular camera undergoing known motion. Such problems arise in machine vision where the position of an object moving in three-dimensional space has to be identified by tracking motion of its projected feature on the two-dimensional image plane. The camera is assumed to be uncalibrated, and an adaptive observer yielding asymptotic estimates of focal length and feature depth is developed that precludes prior knowledge of scene geometry and is simpler than alternative designs. Experimental results using real camera imagery are obtained with the current scheme as well as the extended Kalman filter, and performance of the proposed observer is shown to be better than the extended Kalman filter-based framework.     

基于凸优化改进的相机全局位置估计方法

相机全局位置估计作为运动恢复结构算法（Structure from motion,SfM）中的核心内容一直以来都是计算机视觉领域的研究热点.现有相机全局位置估计方法大多对外点敏感,在处理大规模、无序图像集时表现的尤为明显.增量式SfM中的迭代优化步骤可以剔除大部分的误匹配从而降低外点对估计结果的影响,而全局式SfM中没有有效地剔除误匹配的策略,估计结果受外点影响较大.针对这种情况,本文提出一种改进的相机全局位置估计方法:首先,结合极线约束提出一种新的对误匹配鲁棒的相对平移方向估计算法,减少相对平移方向估计结果中存在的外点;然后,引入平行刚体理论提出一种新的预处理方法将相机全局位置估计转化为一个适定性问题;最后,在此基础上构造了一个对外点鲁棒的凸优化线性估计模型,对模型解算获取相机位置估计全局最优解.本文方法可以很好地融合到当下的全局式SfM流程中.与现有典型方法的对照实验结果表明:在处理大规模、无序图像时,本文方法能显著提高相机全局位置估计的鲁棒性,并保证估计过程的高效性和估计结果的普遍精度.     

基于相机姿势全局运动估计及改进修复算法的视频稳定技术

在视频稳定的过程中,由于摄像机的运动,造成图像的扭曲.针对这种情况,提出一种基于相机姿势的全局运动估计,同时为了克服图像拼接后,部分区域像素丢失的问题,使用改进后调和模型来修复缺少的像素.算法首先提取特征不变量,然后基于这些特征不变量去估计摄像机的运动矢量,相乘各帧间的运动矢量,可以得到每一帧参考第一帧的运动矢量.运用这个矢量可以很好地计算出没有扭曲的图像.运用计算出的图像与视频帧进行拼接,可以很好的解决图像的扭曲的问题.然而,图像拼接完成后可能导致部分区域像素缺少,为了填充缺少像素,算法使用了改进的调和模型来修复缺少区域.实验结果表明,基于相机姿势的全局运动估计可以很好的解决图像扭曲的问题,同时改进的调和模型可以高效的完成对图像的修复.     

Rigid Structure from Motion from a Blind Source Separation Perspective

We present an information theoretic approach to define the problem of structure from motion (SfM) as a blind source separation one. Given that for almost all practical joint densities of shape points, the marginal densities are non-Gaussian, we show how higher-order statistics can be used to provide improvements in shape estimates over the methods of factorization via Singular Value Decomposition (SVD), bundle adjustment and Bayesian approaches. Previous techniques have either explicitly or implicitly used only second-order statistics in models of shape or noise. A further advantage of viewing SfM as a blind source problem is that it easily allows for the inclusion of noise and shape models, resulting in Maximum Likelihood (ML) or Maximum a Posteriori (MAP) shape and motion estimates. A key result is that the blind source separation approach has the ability to recover the motion and shape matrices without the need to explicitly know the motion or shape pdf. We demonstrate that it suffices to know whether the pdf is sub- or super-Gaussian (i.e., semi-parametric estimation) and derive a simple formulation to determine this from the data. We provide extensive experimental results on synthetic and real tracked points in order to quantify the improvement obtained from this technique.     

An Efficient Direct Approach to Visual SLAM

The majority of visual simultaneous localization and mapping (SLAM) approaches consider feature correspondences as an input to the joint process of estimating the camera pose and the scene structure. In this paper, we propose a new approach for simultaneously obtaining the correspondences, the camera pose, the scene structure, and the illumination changes, all directly using image intensities as observations. Exploitation of all possible image information leads to more accurate estimates and avoids the inherent difficulties of reliably associating features. We also show here that, in this case, structural constraints can be enforced within the procedure as well (instead of a posteriori), namely the cheirality, the rigidity, and those related to the lighting variations. We formulate the visual SLAM problem as a nonlinear image alignment task. The proposed parameters to perform this task are optimally computed by an efficient second-order approximation method for fast processing and avoidance of irrelevant minima. Furthermore, a new solution to the visual SLAM initialization problem is described whereby no assumptions are made about either the scene or the camera motion. Experimental results are provided for a variety of scenes, including urban and outdoor ones, under general camera motion and different types of perturbations.      

一种基于尺度不变特征点的运动目标检测算法

针对PTZ（Pan/Tilt/Zoom）摄像机的运动目标检测,提出了一种基于尺度不变（Scale Invariant Feature Transform, SIFT）特征点的运动补偿解决方案．首先,采用相邻两幅图像特征点匹配的方法获得运动补偿,从 而解决了PTZ 摄像机的自主运动带来的问题;然后利用时间差分法得到运动目标的区域;最后利用形态学中 的腐蚀和膨胀方法去除噪声．测试结果验证了算法的实用性和有效性．     

Rank 1 weighted factorization for 3D structure recovery: algorithms and performance analysis

The paper describes the rank 1 weighted factorization solution to the structure from motion problem. This method recovers the 3D structure from the factorization of a data matrix that is rank 1 rather than rank 3. This matrix collects the estimates of the 2D motions of a set of feature points of the rigid object. These estimates are weighted by the inverse of the estimates error standard deviation so that the 2D motion estimates for "sharper" features, which are usually well-estimated, are given more weight, while the noisier motion estimates for "smoother" features are weighted less. We analyze the performance of the rank 1 weighted factorization algorithm to determine what are the most suitable 3D shapes or the best 3D motions to recover the 3D structure of a rigid object from the 2D motions of the features. Our approach is developed for the orthographic camera model. It avoids expensive singular value decompositions by using the power method and is suitable to handle dense sets of feature points and long video sequences. Experimental studies with synthetic and real data illustrate the good performance of our approach.     

A hierarchy of cameras for 3D photography

The view-independent visualization of 3D scenes is most often based on rendering accurate 3D models or utilizes image-based rendering techniques. To compute the 3D structure of a scene from a moving vision sensor or to use image-based rendering approaches, we need to be able to estimate the motion of the sensor from the recorded image information with high accuracy, a problem that has been well-studied. In this work, we investigate the relationship between camera design and our ability to perform accurate 3D photography, by examining the influence of camera design on the estimation of the motion and structure of a scene from video data. By relating the differential structure of the time varying plenoptic function to different known and new camera designs, we can establish a hierarchy of cameras based upon the stability and complexity of the computations necessary to estimate structure and motion. At the low end of this hierarchy is the standard planar pinhole camera for which the structure from motion problem is non-linear and ill-posed. At the high end is a camera, which we call the full field of view polydioptric camera, for which the motion estimation problem can be solved independently of the depth of the scene which leads to fast and robust algorithms for 3D Photography. In between are multiple view cameras with a large field of view which we have built, as well as omni-directional sensors.     

Tracking and Motion Estimation of the Articulated Object: a Hierarchical Kalman Filter Approach

Real-time motion capture plays a very important role in various applications, such as 3D interface for virtual reality systems, digital puppetry, and real-time character animation. In this paper we challenge the problem of estimating and recognizing the motion of articulated objects using theoptical motion capturetechnique. In addition, we present an effective method to control the articulated human figure in realtime.The heart of this problem is the estimation of 3D motion and posture of an articulated, volumetric object using feature points from a sequence of multiple perspective views. Under some moderate assumptions such as smooth motion and known initial posture, we develop a model-based technique for the recovery of the 3D location and motion of a rigid object using a variation of Kalman filter. The posture of the 3D volumatric model is updated by the 2D image flow of the feature points for all views. Two novel concepts – the hierarchical Kalman filter (KHF) and the adaptive hierarchical structure (AHS) incorporating the kinematic properties of the articulated object – are proposed to extend our formulation for the rigid object to the articulated one. Our formulation also allows us to avoid two classic problems in 3D tracking: the multi-view correspondence problem, and the occlusion problem. By adding more cameras and placing them appropriately, our approach can deal with the motion of the object in a very wide area. Furthermore, multiple objects can be handled by managing multiple AHSs and processing multiple HKFs.We show the validity of our approach using the synthetic data acquired simultaneously from the multiple virtual camera in a virtual environment (VE) and real data derived from a moving light display with walking motion. The results confirm that the model-based algorithm works well on the tracking of multiple rigid objects.     

基于DSP的运动目标跟踪系统

描述了一种以TMS320C6701数字信号处理器为核心的高速图像处理板和图像实时采集卡及摄像头构成的实时运动跟踪系统。在对采集的实时图像序列进行如十预处理后．采用了金字塔结构的图像存储方式和特征点跟踪算法埘运动目标进行跟踪．通过对特征点的运算得到目标运动的偏差怍为摄像头运动的参数,是后根据这些参数控制摄像云台持续跟踪运动目标的移动,最后还给出了在复杂背景下跟踪人体的实验结果。     

Automatic geometric reasoning in structure and motion estimation

We present a system for doing automatic surveying or structure and motion analysis given 1D images of a 2D surrounding. Nothing is known about the structure of the scene features or of the motion of the camera. The system automatically identifies and tracks the image of new points and solves the structure and motion problem. One key feature of the system is the ability to hypothesize, test and incorporate simple constraints on the scene, e.g. that two object points are the same, that several points are coplanar. In this paper we develop and test the theory for automatic geometric reasoning. Ideas on hypothesis generation and testing are presented. It is also shown how to update the uncertainty representation of the database.