| ECVision indexed and annotated bibliography of cognitive computer vision publications This bibliography was created by Hilary Buxton and Benoit Gaillard, University of Sussex, as part of ECVision Specific Action 8-1 The complete text version of this BibTeX file is available here: ECVision_bibliography.bib
E. C. {Freeman, W.T.; Haddon, J.A.; Pasztor}
ABSTRACT
We seek a learning-based algorithm that applies to various low-level vision problems. For a given problem, we want to find the scene interpretation that best explains image data. Specializing to the optical flow problem, we may want to infer the projected velocities (scene) which best explain two consecutive image frames (image). We use synthetic data to generate examples of pairs images with their corresponding scene interpretation, the true projected velocities (optical flow). From these data, we learn candidate scene explanations for local image regions, and derive a compatibility function between neighboring scene regions. Given new image data, we propagate beliefs in a Markov network to infer the underlying optical flow. This yields an efficient method to infer low-level scene interpretations. We first present the results of this method for a toy world of irregularly shaped blobs. Then we extend the technique to function on more realistic images, showing reasonable results. |
|
| Site generated on Friday, 06 January 2006
|