Efficient stereo correspondence of uncalibrated images needs the use of complex primitives such a... more Efficient stereo correspondence of uncalibrated images needs the use of complex primitives such as quadrilaterals or perceptual groups. To meet this problem, hierarchic systems are under investigation. The article depicts and comments on one kind of these systems. It deals with contour chains and pertinent points over these chains to reach respectively global and local consistencies through the same generic relaxation technique. Tests are made over several indoor and outdoor real image sequences. They are evaluated and commented. This work is part of the SYMPAS project whose gaol is to achieve 3D planar projective reconstructions from uncalibrated image sequences
Object recognition supported by user interaction for service robots, 2002
ABSTRACT The present paper describes an original technique developed for detecting and matching u... more ABSTRACT The present paper describes an original technique developed for detecting and matching unknown planar structures between a pair of uncalibrated images. Problems related to bad segmentation or to occlusion are automatically solved in most cases. The match process works hierarchically on pertinent point which provide local consistency and on contours which ensure the global consistency. After matching, contours are split into planar contour according to collineations. Tests made over different pairs of images demonstrate the efficiency of our approach.
Proceedings / CVPR, IEEE Computer Society Conference on Computer Vision and Pattern Recognition. IEEE Computer Society Conference on Computer Vision and Pattern Recognition
The planar surfaces (3-D faces) of the objects of a polyhedral scene are built using a trinocular... more The planar surfaces (3-D faces) of the objects of a polyhedral scene are built using a trinocular stereovision system. Before the reconstruction step and after the step of matching, a search is conducted among the set of triplets of matched 2-D faces (supposed to be the projection of 3-D faces) for the triplets which really correspond to planar 3-D surfaces. To accurately reconstruct the surface, a check is conducted in 2-D space for the classes of triplets of faces which correspond to 3-D coplanar structures. It is shown how these 3-D properties are checked in the 2-D space, using tools of projective geometry, allowing the constraint of the 3-D reconstruction in order to maintain these 3-D properties
Proceedings of 1st International Conference on Image Processing, 1994
ABSTRACT We present an approach to construct; 3D facets using trinocular stereovision and charact... more ABSTRACT We present an approach to construct; 3D facets using trinocular stereovision and characterize them by a set of reliable attributes. A facet of a polyedric object is supposed to be projected on the image plane as a homogeneous region surrounded by a chain of 2D segments. These ordered chains, called 2D facets, are extracted from each image of a stereoscopic triplet using the cooperation between single contour map and a series of region maps. These maps are automatically generated according to a coarse-to-fine approach, in order to make the most contrasted regions appear first. The 3D facets are then reconstructed by a technique based on the cooperation of 2D information (2D facets and 2D segments) and 3D information (3D segments provided by a stereovision algorithm). The difficult point is the ordering of the 3D segments constituting the sides of a 3D facet, and is solved by an original algorithm which uses the ordering of three matched 2D facets. Each 3D facet is characterized by a set of intrinsic and extrinsic parameters in order to be efficiently used during the 3D reconstruction of the object to be modelled. As an example, we explain the estimation of the surface of a facet. The original contributions of this work are the cooperation between several heterogeneous data and the use of low quality images, which require robust algorithms
Efficient stereo correspondence of uncalibrated images needs the use of complex primitives such a... more Efficient stereo correspondence of uncalibrated images needs the use of complex primitives such as quadrilaterals or perceptual groups. To meet this problem, hierarchic systems are under investigation. The article depicts and comments on one kind of these systems. It deals with contour chains and pertinent points over these chains to reach respectively global and local consistencies through the same generic relaxation technique. Tests are made over several indoor and outdoor real image sequences. They are evaluated and commented. This work is part of the SYMPAS project whose gaol is to achieve 3D planar projective reconstructions from uncalibrated image sequences
Object recognition supported by user interaction for service robots, 2002
ABSTRACT The present paper describes an original technique developed for detecting and matching u... more ABSTRACT The present paper describes an original technique developed for detecting and matching unknown planar structures between a pair of uncalibrated images. Problems related to bad segmentation or to occlusion are automatically solved in most cases. The match process works hierarchically on pertinent point which provide local consistency and on contours which ensure the global consistency. After matching, contours are split into planar contour according to collineations. Tests made over different pairs of images demonstrate the efficiency of our approach.
Proceedings / CVPR, IEEE Computer Society Conference on Computer Vision and Pattern Recognition. IEEE Computer Society Conference on Computer Vision and Pattern Recognition
The planar surfaces (3-D faces) of the objects of a polyhedral scene are built using a trinocular... more The planar surfaces (3-D faces) of the objects of a polyhedral scene are built using a trinocular stereovision system. Before the reconstruction step and after the step of matching, a search is conducted among the set of triplets of matched 2-D faces (supposed to be the projection of 3-D faces) for the triplets which really correspond to planar 3-D surfaces. To accurately reconstruct the surface, a check is conducted in 2-D space for the classes of triplets of faces which correspond to 3-D coplanar structures. It is shown how these 3-D properties are checked in the 2-D space, using tools of projective geometry, allowing the constraint of the 3-D reconstruction in order to maintain these 3-D properties
Proceedings of 1st International Conference on Image Processing, 1994
ABSTRACT We present an approach to construct; 3D facets using trinocular stereovision and charact... more ABSTRACT We present an approach to construct; 3D facets using trinocular stereovision and characterize them by a set of reliable attributes. A facet of a polyedric object is supposed to be projected on the image plane as a homogeneous region surrounded by a chain of 2D segments. These ordered chains, called 2D facets, are extracted from each image of a stereoscopic triplet using the cooperation between single contour map and a series of region maps. These maps are automatically generated according to a coarse-to-fine approach, in order to make the most contrasted regions appear first. The 3D facets are then reconstructed by a technique based on the cooperation of 2D information (2D facets and 2D segments) and 3D information (3D segments provided by a stereovision algorithm). The difficult point is the ordering of the 3D segments constituting the sides of a 3D facet, and is solved by an original algorithm which uses the ordering of three matched 2D facets. Each 3D facet is characterized by a set of intrinsic and extrinsic parameters in order to be efficiently used during the 3D reconstruction of the object to be modelled. As an example, we explain the estimation of the surface of a facet. The original contributions of this work are the cooperation between several heterogeneous data and the use of low quality images, which require robust algorithms
Uploads