Model selection for two view geometry: a review
Computer vision often involves the estimation of models of the world from visual input. Sometimes it is possible to t several dif-ferent models or hypotheses to a set of data, the choice of exactly which model is usually left to the vision practitioner. This paper explores ways of automating the mod...
| Հիմնական հեղինակ: | |
|---|---|
| Ձևաչափ: | Conference item |
| Լեզու: | English |
| Հրապարակվել է: |
Springer Nature
1999
|
| _version_ | 1826315305380478976 |
|---|---|
| author | Torr, PHS |
| author_facet | Torr, PHS |
| author_sort | Torr, PHS |
| collection | OXFORD |
| description | Computer vision often involves the estimation of models of the world from visual input. Sometimes it is possible to t several dif-ferent models or hypotheses to a set of data, the choice of exactly which model is usually left to the vision practitioner. This paper explores ways of automating the model selection process, with specic emphasis on the least squares problem, and the handling of implicit or nuisance parameters (which in this case equate to 3D structure). The statistical literature is reviewed and it will become apparent that although no one method has yet been developed that will be generally useful for all computer vision problems, there do exist some useful partial solutions. This paper is intended as a pragmatic beginner’s guide to model selection, highlighting the pertinent problems and illustrating them using two view geometry determination. |
| first_indexed | 2024-12-09T03:23:25Z |
| format | Conference item |
| id | oxford-uuid:1186f50a-5a64-4d32-be79-d8f16087d13f |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-12-09T03:23:25Z |
| publishDate | 1999 |
| publisher | Springer Nature |
| record_format | dspace |
| spelling | oxford-uuid:1186f50a-5a64-4d32-be79-d8f16087d13f2024-11-26T14:50:47ZModel selection for two view geometry: a reviewConference itemhttp://purl.org/coar/resource_type/c_5794uuid:1186f50a-5a64-4d32-be79-d8f16087d13fEnglishSymplectic ElementsSpringer Nature1999Torr, PHSComputer vision often involves the estimation of models of the world from visual input. Sometimes it is possible to t several dif-ferent models or hypotheses to a set of data, the choice of exactly which model is usually left to the vision practitioner. This paper explores ways of automating the model selection process, with specic emphasis on the least squares problem, and the handling of implicit or nuisance parameters (which in this case equate to 3D structure). The statistical literature is reviewed and it will become apparent that although no one method has yet been developed that will be generally useful for all computer vision problems, there do exist some useful partial solutions. This paper is intended as a pragmatic beginner’s guide to model selection, highlighting the pertinent problems and illustrating them using two view geometry determination. |
| spellingShingle | Torr, PHS Model selection for two view geometry: a review |
| title | Model selection for two view geometry: a review |
| title_full | Model selection for two view geometry: a review |
| title_fullStr | Model selection for two view geometry: a review |
| title_full_unstemmed | Model selection for two view geometry: a review |
| title_short | Model selection for two view geometry: a review |
| title_sort | model selection for two view geometry a review |
| work_keys_str_mv | AT torrphs modelselectionfortwoviewgeometryareview |