R. Lengagne

École Polytechnique Fédérale de Lausanne, Lausanne, Vaud, Switzerland

Are you R. Lengagne?

Claim your profile

Publications (21)6.75 Total impact

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Given two to four synchronized video streams taken at eye level and from different angles, we show that we can effectively combine a generative model with dynamic programming to accurately follow up to six individuals across thousands of frames in spite of significant occlusions and lighting changes. In addition, we also derive metrically accurate trajectories for each one of them. Our contribution is twofold. First, we demonstrate that our generative model can effectively handle occlusions in each time frame independently, even when the only data available comes from the output of a simple background subtraction algorithm and when the number of individuals is unknown a priori. Second, we show that multi-person tracking can be reliably achieved by processing individual trajectories separately over long sequences, provided that a reasonable heuristic is used to rank these individuals and avoid confusing them with one another.
    IEEE Transactions on Pattern Analysis and Machine Intelligence 03/2008; 30(2):267-82. · 4.80 Impact Factor
  • Source
    F. Fleuret, R. Lengagne, P. Fua
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we show that in a multi-camera context, we can effectively handle occlusions in real-time at each frame independently, even when the only available data comes from the binary output of a simple blob detector, and the number of present individuals is a priori unknown. We start from occupancy probability estimates in a top view and rely on a generative model to yield probability images to be compared with the actual input images. We then refine the estimates so that the probability images match the binary input images as well as possible. We demonstrate the quality of our results on several sequences involving complex occlusions.
    Computer Vision, 2005. ICCV 2005. Tenth IEEE International Conference on; 11/2005
  • Source
    R. Lengagne, P. Fua
    [Show abstract] [Hide abstract]
    ABSTRACT: We propose to incorporate a priori geometric constraints in a 3-D stereo reconstruction scheme to cope with the many cases where image information alone is not sufficient to accurately recover 3-D shape. Our approach is based on the iterative deformation of a 3-D surface mesh to minimize an objective function. We show that combining anisotropic meshing with a nonquadratic approach to regularization enables us to obtain satisfactory reconstruction results using triangulations with few vertices. Structural or numerical constraints can then be added locally to the reconstruction process through a constrained optimization scheme. They improve the reconstruction results and enforce their consistency with a priori knowledge about object shape. The strong description and modeling properties of differential features make them useful tools that can be efficiently used as constraints for 3-D reconstruction
    Computer Vision, 2001. ICCV 2001. Proceedings. Eighth IEEE International Conference on; 02/2001
  • Richard Lengagne, Pascal Fua
    [Show abstract] [Hide abstract]
    ABSTRACT: Conventional stereo algorithms often fail in accurately reconstructing a 3D object because the image data do not provide enough information about the geometry of the object. We propose a way to incorporate a priori information in a reconstruction process from a sequence of calibrated face images. A 3D mesh modeling the face is iteratively deformed in order to minimize an energy function. Differential information extracted from the object shape is used to generate an adaptive mesh. We also propose to explicitly incorporate a priori constraints related to the differential properties of the surface where the image information cannot yield an accurate shape recovery. 1 Introduction 3D face reconstruction is currently receiving a lot of attention in the Computer Vision and Computer Graphics communities. It is a thriving researchfieldwith many applications such as virtual reality, animation, face recognition, etc... In all these cases, the recovered model must be compact and accurate, esp...
    04/2000;
  • R Lengagne, P Fua, O Monga
    [Show abstract] [Hide abstract]
    ABSTRACT: Conventional stereo algorithms often fail in accurately reconstructing a 3D object because the image data do not provide enough information about the geometry of the object. We propose a way to incorporate a priori information in a reconstruction process from a sequence of calibrated face images. A 3D mesh modeling the face is iteratively deformed in order to minimize an energy function. Differential information extracted from the object shape is used to generate an adaptive mesh. We also propose to explicitly incorporate a priori constraints related to the differential properties of the surface where the image information cannot yield an accurate shape recovery.
    Image and Vision Computing 03/2000; · 1.96 Impact Factor
  • Source
    Richard Lengagne, Pascal Fua, Olivier Monga
    Image Vision Comput. 01/2000; 18:337-343.
  • R. Lengagne, P. Fua, O. Monga
    [Show abstract] [Hide abstract]
    ABSTRACT: We propose a way to incorporate a priori information in a 3D stereo reconstruction process from a pair of calibrated face images. A 3D mesh modeling the surface is iteratively deformed in order to minimize an energy function. Differential information about the object shape is used to generate an adaptive mesh that can fulfil the compacity and the accuracy requirements. Moreover in areas where the stereo information is not reliable enough to accurately recover the surface shape, because of inappropriate texture or bad lighting conditions, we incorporate geometric constraints related to the differential properties of the surface, that can be intuitive or refer to predefined geometric properties of the object to be reconstructed. They can be applied to scalar fields, such as curvature values, or structural features, such as crest lines. Therefore, we generate a 3D face model using computer vision techniques that is compact, accurate and consistent with the a priori knowledge about the underlying surface
    Pattern Recognition, 1998. Proceedings. Fourteenth International Conference on; 09/1998
  • Richard Lengagne, Pascal Fua, Olivier Monga
    [Show abstract] [Hide abstract]
    ABSTRACT: . Conventional stereo algorithms often fail in accurately reconstructing a 3D object because the image data do not provide enough information about the geometry of the object. We propose a way to incorporate a priori information in a reconstruction process from a sequence of calibrated face images. A 3D mesh modeling the face is iteratively deformed in order to minimize an energy function in a snake-like process. Differential information about the object shape is used to generate an anisotropic mesh that can both fulfill the compacity and the accuracy requirements. Moreover, in areas where the stereo information is not reliable enough to accurately recover the surface shape, because of inappropriate texture or bad lighting conditions, we propose to incorporate some geometric constraints related to the differential properties of the surface. These constraints can be intuitive or can refer to some predefined geometric properties of the object to be reconstructed. They can be applied to s...
    06/1998;
  • Source
    R. Lengagne, P. Fua, O. Monga
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper proposes a way to incorporate a priori information in a 3D stereo reconstruction process from a pair of calibrated face images. In our framework, a 3D mesh modeling the surface is iteratively deformed in order to minimize an energy function in a snake-like process. Differential information about the object shape as used to generate an anisotropic mesh that can both fulfill the compacity and the accuracy requirements. Moreover, in areas where the stereo information is not reliable enough to accurately recover the surface shape, because of inappropriate texture or bad lighting conditions, we propose to incorporate some geometric constraints related to the differential properties of the surface. These constraints can be intuitive or can refer to some predefined geometric properties of the object to be reconstructed. They can be applied to scalar fields, such as curvature values, or structural features, such as crest lines, governing their location, number, or spatial organization. We demonstrate our approach using faces
    Automatic Face and Gesture Recognition, 1998. Proceedings. Third IEEE International Conference on; 05/1998
  • Richard Lengagne, Olivier Monga, Pascal Fua
    [Show abstract] [Hide abstract]
    ABSTRACT: Stereo reconstruction algorithms often fail to properly deal with complex surfaces, because there is not enough image information. To overcome this problem, we propose to guide the reconstruction process using a priori information about the differential geometry of the object surfaces. We use both linear structures such as crest lines or scalar fields such as curvature values to generate a reconstruction of the surface which is consistent with the differential properties. This method improves the accuracy of the reconstruction around the discontinuities and increases the compactness of the surface representation
    1997 Conference on Computer Vision and Pattern Recognition (CVPR '97), June 17-19, 1997, San Juan, Puerto Rico; 01/1997
  • Source
    R. Lengagne, J.-P. Tarel, O. Monga
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a global scheme for 3D face reconstruction and face segmentation into a limited number of analytical patches from stereo images. From a depth map, we generate a 3D model of the face which is iteratively deformed under stereo and shape-from-shading constraints as well as differential features. This model enables us to improve the quality of the depth map, from which we perform the segmentation and the approximation of the surface
    Automatic Face and Gesture Recognition, 1996., Proceedings of the Second International Conference on; 11/1996
  • R. Lengagne, O. Monga, P. Fua
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we present an approach to reconstruct surfaces from a calibrated pair of images, using stereo information and differential features such as crest lines. It uses an object-centered representation that is optimized to conform to the surface shape. It also extracts typical features such as crest lines and uses them to guide the optimization and the surface reconstruction. We present results using aerial images and terrain data
    Signal Processing, 1996., 3rd International Conference on; 11/1996
  • Source
    R. Lengagne, P. Fua, O. Monga
    [Show abstract] [Hide abstract]
    ABSTRACT: We propose an approach to interleave surface reconstruction from multiple images and feature extraction. It uses an object-centered representation that is optimized to conform to the surface shape. It also extracts typical features such as crest lines and uses them to guide the optimization. We present results using aerial images and terrain data
    Pattern Recognition, 1996., Proceedings of the 13th International Conference on; 09/1996
  • Richard Lengagne, Olivier Monga, Ge Cong, Songde Ma
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we apply successively two methods for corner detection in two-dimensional images, i.e. a differential geometrybased approach relying on multi-scale curvature computation and curvature extrema extraction, and a connexionist approach based on neural networks. We point out the limits of each method and we investigate the way to combine those two strategies in order to get more accurate and more reliable results. This methodology is tested on an indoor scene.
    Image Analysis Applications and Computer Graphics, Third International Computer Science Conference, ICSC'95, Hong Kong, December 11-13, 1995, Proceedings; 01/1995
  • Source
    Olivier Monga, Richard Lengagne, Rachid Deriche
    [Show abstract] [Hide abstract]
    ABSTRACT: : Recently, we have shown that the differential properties of the surfaces represented by 3D volumic images can be recovered using their partial derivatives. For instance, the crest lines can be characterized by the first, second and third partial derivatives of the grey level function I(x; y; z). In this paper, we show that : ffl the computation of the partial derivatives of an image can be improved using recursive filters which approximate the Gaussian filter, ffl a multi-scale approach solves many of the instability problems arising from the computation of the partial derivatives, ffl we illustrate the previous point for the crest line extraction (a crest point is a zerocrossing of the derivative of the maximum curvature along the maximum curvature direction). We present experimental results of crest point extraction on synthetic and 3-D medical data. Key-words: Volume 3D medical images, surface modelling, curvatures, crest lines, multiscale derivation, recursive filtering (R'e...
    12/1994;
  • Source
    O. Monga, R. Lengagne, R. Deriche
    [Show abstract] [Hide abstract]
    ABSTRACT: Presents a multi-scale approach to extract crest lines in volume 3D image. The key point of the authors' approach is to characterize crest points using the first, second and third order partial derivatives of the grey level image function I(x, y, z). These partial derivatives are computed using a recursive filter approximating the Gaussian and its derivatives. Then the width of the filters defines the scale. The authors present experimental results obtained on real data
    Pattern Recognition, 1994. Vol. 1 - Conference A: Computer Vision & Image Processing., Proceedings of the 12th IAPR International Conference on; 11/1994
  • Source
    O. Monga, R. Lengagne, R. Deriche
    [Show abstract] [Hide abstract]
    ABSTRACT: Recently, we have shown that the differential properties of the surfaces represented by 3D volumic images can be recovered using their partial derivatives. For instance, the crest lines can be characterized by the first, second and third partial derivatives of the grey level function I(x, y, z). In this paper, we show that: the computation of the partial derivatives of an image can be improved using recursive filters which approximate the Gaussian filter; a multi-scale approach solves many of the instability problems arising from the computation of the partial derivatives; and we illustrate the previous point for the crest line extraction (a crest point is a zero-crossing of the derivative of the maximum curvature along the maximum curvature direction). We present experimental results of crest point extraction on real 3-D medical data
    Computer Vision and Pattern Recognition, 1994. Proceedings CVPR '94., 1994 IEEE Computer Society Conference on; 07/1994
  • Source
    Olivier Monga, Richard Lengagne, Rachid Deriche, Songde Ma
    Proceedings of IAPR Workshop on Machine Vision Applications, MVA 1994, December 13-15, 1994, Kawasaki, Japan; 01/1994
  • Source
    Francois Fleuret, Richard Lengagne, Pascal Fua
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we show that in a multi-camera context, we can effectively handle occlusions at each time frame independently, even when the only available data comes from the binary output of a fairly primitive motion detector. We start from occupancy probability estimates in a top view and rely on a generative model to yield probability images to be compared with the actual input images. We then refine the estimates so that the probability images match the binary input images as well as possible. We demonstrate the quality of our results on several sequences involving complex occlusions.
  • R. Lengagne, O. Monga, P. Fua