Results 1  10
of
12
Correspondencefree synchronization and reconstruction in a nonrigid scene
 In Proc. Workshop on Vision and Modelling of Dynamic Scenes
, 2002
"... 3D reconstruction of a dynamic nonrigid scene from features in two cameras usually requires synchronization and correspondences between the cameras. These may be hard to achieve due to occlusions, wide baseline, different zoom scales, etc. In this work we present an algorithm for reconstructing a ..."
Abstract

Cited by 21 (1 self)
 Add to MetaCart
3D reconstruction of a dynamic nonrigid scene from features in two cameras usually requires synchronization and correspondences between the cameras. These may be hard to achieve due to occlusions, wide baseline, different zoom scales, etc. In this work we present an algorithm for reconstructing a dynamic scene from sequences acquired by two uncalibrated nonsynchronized fixed affine cameras. It is assumed that (possibly) different points are tracked in the two sequences. The only constraint used to relate the two cameras is that every 3D point tracked in one sequence can be described as a linear combination of some of the 3D points tracked in the other sequence. Such constraint is useful, for example, for articulated objects. We may track some points on an arm in the first sequence, and some other points on the same arm in the second sequence. On the other extreme, this model can be used for generally moving points tracked in both sequences without knowing the correct permutation. In between, this model can cover nonrigid bodies, with local rigidity constraints. 1 We present linear algorithms for synchronizing the two sequences and reconstructing the 3D points tracked in both views. Outlier points are automatically detected and discarded. The algorithm can handle both 3D objects and planar objects in a unified framework, therefore avoiding numerical problems existing in other methods. 1
Variational stereovision and 3d scene flow estimation with statistical similarity measures
 IEEE International Conference on Computer Vision
, 2003
"... We present a common variational framework for dense depth recovery and dense threedimensional motion field estimation from multiple video sequences, which is robust to camera spectral sensitivity differences and illumination changes. For this purpose, we first show that both problems reduce to a ge ..."
Abstract

Cited by 20 (4 self)
 Add to MetaCart
(Show Context)
We present a common variational framework for dense depth recovery and dense threedimensional motion field estimation from multiple video sequences, which is robust to camera spectral sensitivity differences and illumination changes. For this purpose, we first show that both problems reduce to a generic image matching problem after backprojecting the input images onto suitable surfaces. We then solve this matching problem in the case of statistical similarity criteria that can handle frequently occurring nonaffine image intensities dependencies. Our method leads to an efficient and elegant implementation based on fast recursive filters. We obtain good results on real images. 1.
Motion  Stereo Integration for Depth Estimation
 In Eur. Conf. on Computer Vision
, 2002
"... Depth extraction with a mobile stereo system is described. ..."
Abstract

Cited by 17 (1 self)
 Add to MetaCart
Depth extraction with a mobile stereo system is described.
Modelling nonrigid dynamic scenes from multiview image sequences
 The Handbook of Mathematical Models in Computer Vision, chapter 27
, 2005
"... ABSTRACT This chapter focuses on the problem of obtaining a complete spatiotemporal description of some objects undergoing a nonrigid motion, given several calibrated and synchronized videos of the scene. Using stereovision and scene flow methods in conjunction, the threedimensional shape and the ..."
Abstract

Cited by 5 (1 self)
 Add to MetaCart
(Show Context)
ABSTRACT This chapter focuses on the problem of obtaining a complete spatiotemporal description of some objects undergoing a nonrigid motion, given several calibrated and synchronized videos of the scene. Using stereovision and scene flow methods in conjunction, the threedimensional shape and the nonrigid threedimensional motion field of the objects can be recovered. We review the unrealistic photometric and geometric assumptions which plague existing methods. A novel method based on deformable surfaces is proposed to alleviate some of these limitations. 1
Integrating Stereo Disparity and Optical Flow by Closelycoupled Method
, 2011
"... As a convergence method for stereo matching and motion estimation, this paper presents an equation, called DisparityOptical flow Equation, that relates disparity with optical flow in rectified images. Considering this equation as a constraint, this paper suggests an algorithm, called Simultaneous ..."
Abstract
 Add to MetaCart
(Show Context)
As a convergence method for stereo matching and motion estimation, this paper presents an equation, called DisparityOptical flow Equation, that relates disparity with optical flow in rectified images. Considering this equation as a constraint, this paper suggests an algorithm, called Simultaneous Disparity and Optical Flow under Epipolar Constraints, that efficiently determines the disparity and optical flow. This algorithm is completely different from the previous approaches that try to compromise the outputs of the two modules. The experiments show that the algorithm can resolve the ambiguous situations in which either stereo matching or motion estimation fail to yield satisfactory results.
Geometric Relationship between Stereo Disparity and Optical Flow and an Efficient Recursive Algorithm
"... We suggest a relationship, called stereomotion equation, between stereo disparity and optical flow, and a recursive filter, as an efficient algorithm to estimate the two quantities. We show that close spatial and temporal relationships exist between the two quantities. The importance of this discov ..."
Abstract
 Add to MetaCart
(Show Context)
We suggest a relationship, called stereomotion equation, between stereo disparity and optical flow, and a recursive filter, as an efficient algorithm to estimate the two quantities. We show that close spatial and temporal relationships exist between the two quantities. The importance of this discovery is that the vision quantities can be computed simultaneously, on the image level, unlike previous approaches that tried to determine the two quantities separately. This algorithm is general because it can be reduced to the separate methods to estimate either disparity or optical flow. Because the two vision quantities help mutually for the best matches, the results tend to be more accurate and reliable than the separate methods. As an efficient algorithm, we suggest a recursive filter in which the two vision quantities are determined alternatively by time and measurement updates. This algorithm was tested on both synthetic and natural scenes containing moving objects and performed better than separate calculation of disparity and optical flow.
1. Stereo, motion and structure 14 Nonrigid Stereomotion
"... Using a calibrated stereo pair is a common and practical solution to obtain reliable 3D reconstructions. In its simpler formulation, once the stereo rig is calibrated, the depth of points in the image is estimated by applying triangulation (Trucco & Verri, 1998). In order to obtain accurate dep ..."
Abstract
 Add to MetaCart
(Show Context)
Using a calibrated stereo pair is a common and practical solution to obtain reliable 3D reconstructions. In its simpler formulation, once the stereo rig is calibrated, the depth of points in the image is estimated by applying triangulation (Trucco & Verri, 1998). In order to obtain accurate depth estimates, the cameras are usually separated from each other by a significant baseline thus creating widely spaced observations of the same object. The disadvantage of this configuration though, is that having a wide baseline makes the matching of features between pairs of views a more challenging problem. On the other hand, the task of computing temporal tracks from single camera sequences is relatively easier since the images are closely spaced in time. As a drawback, disparities may be insufficient to obtain a reliable depth estimation and, as a result, longer sequences are needed to infer the 3D structure. Particularly, in the case of nonrigid structure, a sufficient overall rigid motion is necessary to allow the algorithms to estimate the reconstruction parameters correctly. Hence, a question of relevant interest is the feasibility of an approach that efficiently fuses the positive aspects of both methods. The problem of recovering 3D structure using a
Factorizationbased Nonrigid Shape Modeling and Tracking in StereoMotion
"... In recent years, researchers have tackled the topic of augmenting “structure from motion ” with stereo information. Unfortunately nearly all stereomotion algorithms assume that the scene is rigid. In [16] the factorizationbased structurefrommotion method for rigid objects was proposed, and soon i ..."
Abstract
 Add to MetaCart
(Show Context)
In recent years, researchers have tackled the topic of augmenting “structure from motion ” with stereo information. Unfortunately nearly all stereomotion algorithms assume that the scene is rigid. In [16] the factorizationbased structurefrommotion method for rigid objects was proposed, and soon it was extended to nonrigid or deformable objects in [17]. In this paper we propose a framework of factorizationbased nonrigid shape modeling and tracking in stereomotion. We construct a measurement matrix consisting of the stereomotion data captured from a stereorig. Organized in a particular way this matrix could be factored by Singular Value Decomposition (SVD) to get the 3D basis shapes, their configuration weights, rigid motion and camera geometry. With this, the stereo correspondences can be inferred from motion correspondences only requiring that a minimum of 3K point stereo correspondences (where K is the dimension of shape basis space) are determined in advance. Basically this framework still keeps the usage potential of rank constraints, meanwhile it has other advantages such as simpler correspondence and accurate reconstruction even with short image sequences. Results of synthesized data and real stereo sequences are given to demonstrate the performance. 1.