Parsimonious labeling
We propose a new family of discrete energy minimization problems, which we call parsimonious labeling. Our energy function consists of unary potentials and high-order clique potentials. While the unary potentials are arbitrary, the clique potentials are proportional to the diversity of the set of un...
| Main Authors: | , |
|---|---|
| Format: | Conference item |
| Published: |
Institute of Electrical and Electronics Engineers
2016
|
| _version_ | 1826271547944337408 |
|---|---|
| author | Dokania, P Mudigonda, P |
| author_facet | Dokania, P Mudigonda, P |
| author_sort | Dokania, P |
| collection | OXFORD |
| description | We propose a new family of discrete energy minimization problems, which we call parsimonious labeling. Our energy function consists of unary potentials and high-order clique potentials. While the unary potentials are arbitrary, the clique potentials are proportional to the diversity of the set of unique labels assigned to the clique. Intuitively, our energy function encourages the labeling to be parsimonious, that is, use as few labels as possible. This in turn allows us to capture useful cues for important computer vision applications such as stereo correspondence and image denoising. Furthermore, we propose an efficient graph-cuts based algorithm for the parsimonious labeling problem that provides strong theoretical guarantees on the quality of the solution. Our algorithm consists of three steps. First, we approximate a given diversity using a mixture of a novel hierarchical Pn Potts model. Second, we use a divide-andconquer approach for each mixture component, where each subproblem is solved using an efficient expansion algorithm. This provides us with a small number of putative labelings, one for each mixture component. Third, we choose the best putative labeling in terms of the energy value. Using both synthetic and standard real datasets, we show that our algorithm significantly outperforms other graph-cuts based approaches. |
| first_indexed | 2024-03-06T21:58:23Z |
| format | Conference item |
| id | oxford-uuid:4dbd5266-b883-4abb-acff-5017e97f61f2 |
| institution | University of Oxford |
| last_indexed | 2024-03-06T21:58:23Z |
| publishDate | 2016 |
| publisher | Institute of Electrical and Electronics Engineers |
| record_format | dspace |
| spelling | oxford-uuid:4dbd5266-b883-4abb-acff-5017e97f61f22022-03-26T15:57:06ZParsimonious labelingConference itemhttp://purl.org/coar/resource_type/c_5794uuid:4dbd5266-b883-4abb-acff-5017e97f61f2Symplectic Elements at OxfordInstitute of Electrical and Electronics Engineers2016Dokania, PMudigonda, PWe propose a new family of discrete energy minimization problems, which we call parsimonious labeling. Our energy function consists of unary potentials and high-order clique potentials. While the unary potentials are arbitrary, the clique potentials are proportional to the diversity of the set of unique labels assigned to the clique. Intuitively, our energy function encourages the labeling to be parsimonious, that is, use as few labels as possible. This in turn allows us to capture useful cues for important computer vision applications such as stereo correspondence and image denoising. Furthermore, we propose an efficient graph-cuts based algorithm for the parsimonious labeling problem that provides strong theoretical guarantees on the quality of the solution. Our algorithm consists of three steps. First, we approximate a given diversity using a mixture of a novel hierarchical Pn Potts model. Second, we use a divide-andconquer approach for each mixture component, where each subproblem is solved using an efficient expansion algorithm. This provides us with a small number of putative labelings, one for each mixture component. Third, we choose the best putative labeling in terms of the energy value. Using both synthetic and standard real datasets, we show that our algorithm significantly outperforms other graph-cuts based approaches. |
| spellingShingle | Dokania, P Mudigonda, P Parsimonious labeling |
| title | Parsimonious labeling |
| title_full | Parsimonious labeling |
| title_fullStr | Parsimonious labeling |
| title_full_unstemmed | Parsimonious labeling |
| title_short | Parsimonious labeling |
| title_sort | parsimonious labeling |
| work_keys_str_mv | AT dokaniap parsimoniouslabeling AT mudigondap parsimoniouslabeling |