A bottom up algebraic approach to motion segmentation

Authors:
Dheeraj Singaraju;René Vidal
Affiliations:
Center for Imaging Science, Johns Hopkins University, Baltimore, MD;Center for Imaging Science, Johns Hopkins University, Baltimore, MD
Venue:
ACCV'06 Proceedings of the 7th Asian conference on Computer Vision - Volume Part I
Year:
2006

Citing 7
Cited 2

An Integrated Bayesian Approach to Layer Extraction from Image Sequences

IEEE Transactions on Pattern Analysis and Machine Intelligence
Smoothness in Layers: Motion segmentation using nonparametric mixture estimation.

CVPR '97 Proceedings of the 1997 Conference on Computer Vision and Pattern Recognition (CVPR '97)
A Unified Mixture Framework for Motion Segmentation: Incorporating Spatial Coherence and Estimating the Number of Models

CVPR '96 Proceedings of the 1996 Conference on Computer Vision and Pattern Recognition (CVPR '96)
Segmentation of Dynamic Scenes from Image Intensities

MOTION '02 Proceedings of the Workshop on Motion and Video Computing
Layered representation of motion video using robust maximum-likelihood estimation of mixture models and MDL encoding

ICCV '95 Proceedings of the Fifth International Conference on Computer Vision
Motion Competition: A Variational Approach to Piecewise Parametric Motion Segmentation

International Journal of Computer Vision
A Closed Form Solution to Direct Motion Segmentation

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02

3D motion segmentation from straight-line optical flow

MCAM'07 Proceedings of the 2007 international conference on Multimedia content analysis and mining
Direct segmentation of multiple 2-D motion models of different types

WDV'05/WDV'06/ICCV'05/ECCV'06 Proceedings of the 2005/2006 international conference on Dynamical vision

Quantified Score

Hi-index	0.00

Visualization

Abstract

We present a bottom up algebraic approach for segmenting multiple 2D motion models directly from the partial derivatives of an image sequence. Our method fits a polynomial called the multibody brightness constancy constraint (MBCC) to a window around each pixel of the scene and obtains a local motion model from the derivatives of the MBCC. These local models are then clustered to obtain the parameters of the motion models for the entire scene. Motion segmentation is obtained by assigning to each pixel the dominant motion model in a window around it. Our approach requires no initialization, can handle multiple motions in a window (thus dealing with the aperture problem) and automatically incorporates spatial regularization. Therefore, it naturally combines the advantages of both local and global approaches to motion segmentation. Experiments on real data compare our method with previous local and global approaches.