Robust higher order potentials for enforcing label consistency
This paper proposes a novel framework for labelling problems which is able to combine multiple segmentations in a principled manner. Our method is based on higher order conditional random fields and uses potentials defined on sets of pixels (image segments) generated using unsupervised segmentation...
| Main Authors: | , , |
|---|---|
| Format: | Journal article |
| Language: | English |
| Published: |
Springer
2009
|
| _version_ | 1826313439312609280 |
|---|---|
| author | Kohli, P Ladický, L Torr, PHS |
| author_facet | Kohli, P Ladický, L Torr, PHS |
| author_sort | Kohli, P |
| collection | OXFORD |
| description | This paper proposes a novel framework for labelling problems which is able to combine multiple segmentations in a principled manner. Our method is based on higher order conditional random fields and uses potentials defined on sets of pixels (image segments) generated using unsupervised segmentation algorithms. These potentials enforce label consistency in image regions and can be seen as a generalization of the commonly used pairwise contrast sensitive smoothness potentials. The higher order potential functions used in our framework take the form of the Robust P n model and are more general than the P n Potts model recently proposed by Kohli et al. We prove that the optimal swap and expansion moves for energy functions composed of these potentials can be computed by solving a st-mincut problem. This enables the use of powerful graph cut based move making algorithms for performing inference in the framework. We test our method on the problem of multi-class object segmentation by augmenting the conventional crf used for object segmentation with higher order potentials defined on image regions. Experiments on challenging data sets show that integration of higher order potentials quantitatively and qualitatively improves results leading to much better definition of object boundaries. We believe that this method can be used to yield similar improvements for many other labelling problems. © 2009 Springer Science+Business Media, LLC. |
| first_indexed | 2024-09-25T04:14:43Z |
| format | Journal article |
| id | oxford-uuid:f5b8b0ce-cb7d-4c41-b10a-dbdc03d813df |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-09-25T04:14:43Z |
| publishDate | 2009 |
| publisher | Springer |
| record_format | dspace |
| spelling | oxford-uuid:f5b8b0ce-cb7d-4c41-b10a-dbdc03d813df2024-07-08T15:46:48ZRobust higher order potentials for enforcing label consistencyJournal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:f5b8b0ce-cb7d-4c41-b10a-dbdc03d813dfEnglishSymplectic ElementsSpringer2009Kohli, PLadický, LTorr, PHSThis paper proposes a novel framework for labelling problems which is able to combine multiple segmentations in a principled manner. Our method is based on higher order conditional random fields and uses potentials defined on sets of pixels (image segments) generated using unsupervised segmentation algorithms. These potentials enforce label consistency in image regions and can be seen as a generalization of the commonly used pairwise contrast sensitive smoothness potentials. The higher order potential functions used in our framework take the form of the Robust P n model and are more general than the P n Potts model recently proposed by Kohli et al. We prove that the optimal swap and expansion moves for energy functions composed of these potentials can be computed by solving a st-mincut problem. This enables the use of powerful graph cut based move making algorithms for performing inference in the framework. We test our method on the problem of multi-class object segmentation by augmenting the conventional crf used for object segmentation with higher order potentials defined on image regions. Experiments on challenging data sets show that integration of higher order potentials quantitatively and qualitatively improves results leading to much better definition of object boundaries. We believe that this method can be used to yield similar improvements for many other labelling problems. © 2009 Springer Science+Business Media, LLC. |
| spellingShingle | Kohli, P Ladický, L Torr, PHS Robust higher order potentials for enforcing label consistency |
| title | Robust higher order potentials for enforcing label consistency |
| title_full | Robust higher order potentials for enforcing label consistency |
| title_fullStr | Robust higher order potentials for enforcing label consistency |
| title_full_unstemmed | Robust higher order potentials for enforcing label consistency |
| title_short | Robust higher order potentials for enforcing label consistency |
| title_sort | robust higher order potentials for enforcing label consistency |
| work_keys_str_mv | AT kohlip robusthigherorderpotentialsforenforcinglabelconsistency AT ladickyl robusthigherorderpotentialsforenforcinglabelconsistency AT torrphs robusthigherorderpotentialsforenforcinglabelconsistency |