WEKO3
アイテム
Surface Color Estimation of Large Scale Diffuse Objects under Outdoor Environment
https://doi.org/10.15083/00002415
https://doi.org/10.15083/0000241552ca9cc5-b336-4164-989c-08020fc4d9a8
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
48057404.pdf (8.6 MB)
|
|
| Item type | 学位論文 / Thesis or Dissertation(1) | |||||
|---|---|---|---|---|---|---|
| 公開日 | 2012-03-01 | |||||
| タイトル | ||||||
| タイトル | Surface Color Estimation of Large Scale Diffuse Objects under Outdoor Environment | |||||
| 言語 | ||||||
| 言語 | eng | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | surface color | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | surface reflectance | |||||
| キーワード | ||||||
| 主題Scheme | Other | |||||
| 主題 | color constancy | |||||
| 資源タイプ | ||||||
| 資源 | http://purl.org/coar/resource_type/c_46ec | |||||
| タイプ | thesis | |||||
| ID登録 | ||||||
| ID登録 | 10.15083/00002415 | |||||
| ID登録タイプ | JaLC | |||||
| その他のタイトル | ||||||
| その他のタイトル | 屋外環境下における大規模拡散反射物体の表面色推定 | |||||
| 著者 |
KAWAKAMI, REI
× KAWAKAMI, REI |
|||||
| 著者別名 | ||||||
| 識別子Scheme | WEKO | |||||
| 識別子 | 6678 | |||||
| 姓名 | 川上, 玲 | |||||
| 著者所属 | ||||||
| 著者所属 | 大学院情報理工学系研究科電子情報学専攻 | |||||
| 著者所属 | ||||||
| 著者所属 | Graduate School of Information Science and Technology Department of Information and Communication Engineering The University of Tokyo | |||||
| Abstract | ||||||
| 内容記述タイプ | Abstract | |||||
| 内容記述 | Digital three-dimensional models created by computer vision and graphics techniques are becoming widely used for a variety of purposes. Specifically, modeling cultural heritage objects has attracted considerable attention, since such objects are worth preserving, and the data can be utilized for restoration when an object faces the crisis of collapse. Automation for creating 3D models has therefore attracted much interest, since most models are currently created by manual operation, adding significantly to the cost. Creating an accurate model of an object requires knowledge of the object's shape and surface reflectance. Acquiring shape information is facilitated by the development of sensors and the progress of data processing algorithms, but acquiring surface reflectance properties remains a challenge, specifically with outdoor objects. This paper targets large-scale objects such as architectural structures in an outdoor environment. The size of target objects may be as much as 100 m by 100 m by 50 m. Measuring the surface properties of such huge objects is a challenge. The appearance of an object can be modeled by mapping image textures to the known shape of the object. However, to achieve consistent colors among image textures, the effect of illumination has to be removed before mapping these textures by using surface color estimation and surface reflectance estimation techniques. Two methods that calculate a surface color by a pixel-based operation are presented. Most previous methods assume uniform illumination in a scene, but this is not always true in images with shadows or with curved objects. The proposed methods enable pixel-based operation by utilizing illumination change. Two models of illumination colors that we introduce enable a surface color to be uniquely determined from two pixel values. First, the paper proposes a method that uses blackbody radiation and analyzes the stability and practicality of the method. Then, a more practical method is proposed that can perform robust estimation using a statistical model derived from outdoor illumination data. Robust estimation is achieved by introducing the plausible range of outdoor illumination colors. In practical situation, surface reflectance would be required for relighting purposes. A method is presented to estimate surface reflectance from spherical images with known shape information. Spherical images have nearly a 360-degree field of view; they capture target objects and surrounding illumination at one shot. Therefore they do not require specific apparatus or calibration of exposure times, apertures, and camera gain factors. Furthermore, geometric calibration between an image and shape information becomes robust owing to the characteristic of a spherical camera. Measurement and data-processing cost will be decreased by the method compared to previous methods that need elaborate procedures. This is critical specifically for large-scale objects. The main contribution of this thesis is that the author has proposed three methods that estimate surface properties of an object. It can be summarized by the three following points: First, the research provides insights into the stability and practicality of pixel-based surface color estimation. Second, a pixel-based method for surface color estimation has been developed that is robust and accurate even for real outdoor objects. None of the conventional methods can perform a pixel-based operation with higher accuracy than the proposed method. Third, an efficient method has been developed that estimates surface reflectance of large-scale objects under outdoor environment. The proposed techniques form the foundation for developing a system that models the appearance of a large-scale object in an outdoor environment. | |||||
| 書誌情報 | 発行日 2008-03 | |||||
| 日本十進分類法 | ||||||
| 主題Scheme | NDC | |||||
| 主題 | 548 | |||||
| 学位名 | ||||||
| 学位名 | 博士(情報理工学) | |||||
| 学位 | ||||||
| 値 | doctoral | |||||
| 学位分野 | ||||||
| Information Science and Technology (情報理工学) | ||||||
| 学位授与機関 | ||||||
| 学位授与機関名 | University of Tokyo (東京大学) | |||||
| 研究科・専攻 | ||||||
| Department of Information and Communication Engineering, Graduate School of Information Science and Technology (情報理工学系研究科電子情報学専攻) | ||||||
| 学位授与年月日 | ||||||
| 学位授与年月日 | 2008-03-24 | |||||
| 学位授与番号 | ||||||
| 学位授与番号 | 甲第23943号 | |||||
| 学位記番号 | ||||||
| 博情第188号 | ||||||
