| Item type |
SIG Technical Reports(1) |
| 公開日 |
2015-05-11 |
| タイトル |
|
|
タイトル |
Shape from Scattering: Shape Estimation for Translucent Objects Based on Light Transport Analysis |
| タイトル |
|
|
言語 |
en |
|
タイトル |
Shape from Scattering: Shape Estimation for Translucent Objects Based on Light Transport Analysis |
| 言語 |
|
|
言語 |
eng |
| 資源タイプ |
|
|
資源タイプ識別子 |
http://purl.org/coar/resource_type/c_18gh |
|
資源タイプ |
technical report |
| 著者所属 |
|
|
|
Osaka University |
| 著者所属 |
|
|
|
Nara Institute of Science and Technology |
| 著者所属 |
|
|
|
Osaka University |
| 著者所属 |
|
|
|
Osaka University |
| 著者所属(英) |
|
|
|
en |
|
|
Osaka University |
| 著者所属(英) |
|
|
|
en |
|
|
Nara Institute of Science and Technology |
| 著者所属(英) |
|
|
|
en |
|
|
Osaka University |
| 著者所属(英) |
|
|
|
en |
|
|
Osaka University |
| 著者名 |
Chika, Inoshita
Yasuhiro, Mukaigawa
Yasuyuki, Matsushita
Yasushi, Yagi
|
| 著者名(英) |
Chika, Inoshita
Yasuhiro, Mukaigawa
Yasuyuki, Matsushita
Yasushi, Yagi
|
| 論文抄録 |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Accurate shape measurement has a long history in computer vision yet still remains a challenging task. In particular, a translucent object is one of the most difficult targets because of complex scattering and transmission of light in the media, which cannot be represented by a simple image formation model. To address this issue, we propose a shape from scattering framework, which enables estimation of 3D shape of translucent objects by either exploiting or explicitly modeling scattering. Our key idea is to represent the inherently complex relationship between the object shape and scene appearance due to scattering using a simple yet accurate model. Because the scattering distribution depends on optical properties of target material, we first measure and analyze a light transport on various types of translucent material as 8-dimensional bidirectional scattering surface reflectance distribution function (8D BSSRDF), then determine appropriate parameters of a scattering model. Based on the analysis of BSSRDF, we estimate object shape from observed appearances with a light attenuation model of single scattering for optically thin material and convolutional scattering distribution model for optically thick material, respectively. We show the effectiveness of our shape estimation method with experiments using translucent objects in real scenes. Although scattering has been regarded as a nuisance and eliminated by employing various pre-processing methods in the past, our shape from scattering framework eliminates the need of such external methods and makes direct inference of object shape in the presence of scattering. |
| 論文抄録(英) |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Accurate shape measurement has a long history in computer vision yet still remains a challenging task. In particular, a translucent object is one of the most difficult targets because of complex scattering and transmission of light in the media, which cannot be represented by a simple image formation model. To address this issue, we propose a shape from scattering framework, which enables estimation of 3D shape of translucent objects by either exploiting or explicitly modeling scattering. Our key idea is to represent the inherently complex relationship between the object shape and scene appearance due to scattering using a simple yet accurate model. Because the scattering distribution depends on optical properties of target material, we first measure and analyze a light transport on various types of translucent material as 8-dimensional bidirectional scattering surface reflectance distribution function (8D BSSRDF), then determine appropriate parameters of a scattering model. Based on the analysis of BSSRDF, we estimate object shape from observed appearances with a light attenuation model of single scattering for optically thin material and convolutional scattering distribution model for optically thick material, respectively. We show the effectiveness of our shape estimation method with experiments using translucent objects in real scenes. Although scattering has been regarded as a nuisance and eliminated by employing various pre-processing methods in the past, our shape from scattering framework eliminates the need of such external methods and makes direct inference of object shape in the presence of scattering. |
| 書誌レコードID |
|
|
収録物識別子タイプ |
NCID |
|
収録物識別子 |
AA11131797 |
| 書誌情報 |
研究報告コンピュータビジョンとイメージメディア（CVIM）
巻 2015-CVIM-197,
号 34,
p. 1-16,
発行日 2015-05-11
|
| ISSN |
|
|
収録物識別子タイプ |
ISSN |
|
収録物識別子 |
2188-8701 |
| Notice |
|
|
|
SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc. |
| 出版者 |
|
|
言語 |
ja |
|
出版者 |
情報処理学会 |
IPSJ-CVIM15197034.pdf (11.3 MB)
