Scene Segmentation from Visual Motion Using Global Optimization
IEEE Transactions on Pattern Analysis and Machine Intelligence
Optical Flow Estimation: An Error Analysis of Gradient-Based Methods with Local Optimization
IEEE Transactions on Pattern Analysis and Machine Intelligence
Visual reconstruction
Image Flow Segmentation and Estimation by Constraint Line Clustering
IEEE Transactions on Pattern Analysis and Machine Intelligence
Computation of component image velocity from local phase information
International Journal of Computer Vision
Bayesian Estimation of Motion Vector Fields
IEEE Transactions on Pattern Analysis and Machine Intelligence
Qualitative constraints for structure-from-motion
CVGIP: Image Understanding - Special issue on purposive, qualitative, active vision
Performance of optical flow techniques
International Journal of Computer Vision
Optical flow estimation and the interaction between measurement errors at adjacent pixel positions
International Journal of Computer Vision
Local and Global Interpretation of Moving Images
Local and Global Interpretation of Moving Images
Measurement of Visual Motion
Multimodal Estimation of Discontinuous Optical Flow using Markov Random Fields
IEEE Transactions on Pattern Analysis and Machine Intelligence
A Stereo Matching Algorithm with an Adaptive Window: Theory and Experiment
IEEE Transactions on Pattern Analysis and Machine Intelligence
Particle Video: Long-Range Motion Estimation Using Point Trajectories
International Journal of Computer Vision
Optical-flow based on an edge-avoidance procedure
Computer Vision and Image Understanding
Optical flow estimation for motion-compensated compression
Image and Vision Computing
Hi-index | 0.00 |
Most optical flow estimation techniques have substantial difficultiesdealing with flow discontinuities. Methods which simultaneously detectflow boundaries and use the detected boundaries to aid in flowestimation can produce significantly improved results. Currentapproaches to implementing these methods still have importantlimitations, however. We demonstrate three such problems: errors dueto the mixture of image properties across boundaries, an intrinsicambiguity in boundary location when only short sequences areconsidered, and difficulties insuring that the motion of a boundaryaids in flow estimation for the surface to which it is attached withoutcorrupting the flow estimates for the occluded surface on the otherside. The first problem can be fixed by basing flow estimation only onimage changes at edges. The second requires an analysis of longer timeintervals. The third can be aided by using a boundary detectionmechanism which classifies the sides of boundaries as occluding andoccluded at the same time as the boundaries are detected.