On Kineopsis and Computation of Structure and Motion
IEEE Transactions on Pattern Analysis and Machine Intelligence
From image sequences towards conceptual descriptions
Image and Vision Computing
A Projection Operator for the Restoration of Divergence-Free Vector Fields
IEEE Transactions on Pattern Analysis and Machine Intelligence
Interpretation of Image Flow: A Spatio-Temporal Approach
IEEE Transactions on Pattern Analysis and Machine Intelligence
Inherent Ambiguities in Recovering 3-D Motion and Structure from a Noisy Flow Field
IEEE Transactions on Pattern Analysis and Machine Intelligence
Motion Field and Optical Flow: Qualitative Properties
IEEE Transactions on Pattern Analysis and Machine Intelligence
Obstacle Avoidance Using Flow Field Divergence
IEEE Transactions on Pattern Analysis and Machine Intelligence
The analytic structure of image flows: deformation and segmentation
Computer Vision, Graphics, and Image Processing
Bounds on time-to-collision and rotational component from first-order derivatives of image flow
Computer Vision, Graphics, and Image Processing
Qualitative motion understanding
IJCAI'87 Proceedings of the 10th international joint conference on Artificial intelligence - Volume 2
IEEE Transactions on Pattern Analysis and Machine Intelligence
IEEE Transactions on Pattern Analysis and Machine Intelligence
Qualitative and Quantitative Car Tracking from a Range Image Sequence
CVPR '98 Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
| Hi-index | 0.00 |
A qualitative approach to get, information about 3D kinematic behaviour of objects in a scene from apparent motion in the image sequence is presented. The process is in fact two-fold: first partition of the image into areas comprising a unique motion; and second, symbolic description of the motion of each area. Here the second part is discussed. Kinematic description involves motion and trajectory type. It relies on geometrical cues tied to the velocity field such as divergence or rotational terms. First, it is shown how a complete set of such cues can be derived through a first order development of the 2D velocity field. Second, the relation between these cues and the 3D motion parameters is established; which allows the determination of a set of labels associated with different kinematic configurations, Third, the label validation step is solved using a statistical approach (in fact, two methods have been studied). This approach avoids determining explicit 3D parameters such as a depth snap or 3D motion measurements. Several experiments on different sequences are reported.