自由視点映像における背景モデル作成のための一手法 A Room Modeling Method for 3D Video

この論文にアクセスする

この論文をさがす

著者

抄録

自由視点映像における背景領域の形状モデルを簡便に生成する手法を提案する.前景領域の自由視点映像は,視体積交差法などの手法により 3 次元形状を復元することで実現されている.一方で,背景領域の形状モデルは CAD などの 3D モデリングソフトを用いて作成されることが多い.この場合,詳細なモデルの作成には手間がかかり,また,簡易的に作成した単純な形状のモデルを使用すると映像が不自然なものとなっていた.本提案手法では,背景領域の形状を複数の平面で近似することとし,多視点画像に写り込んだ平面の頂点を対応点として指定するという簡単な処理により背景モデルを生成する.各対応点の 3 次元座標は,前景領域の 3 次元推定に用いられる多視点カメラのキャリブレーション情報を基にステレオ処理により算出される.実際の形状に比較的近いモデルを簡易に生成することが可能である.This paper proposes a method to reconstruct the 3D shape model of background regionssimply for generating a 3D video. As the progress of computer vision technologies continues, many 3D video systems have been developed that tend to focus on automatic 3D reconstruction to capture target (foreground) objects. On the other hand, the 3D shapes of background regions such as walls and floors are generally made manually using 3D modeling software, i.e., CAD, or as a textured cube that just assumes the room. If we make background models using 3D modeling software, the quality is very high but it costs a lot of time and money. On the other hand, just making a cube is too simple to represent the outer space with sufficient appearance. To solve this problem, we propose a method to reconstruct the background model with simple operations but whose quality remains higher than cubes. In our method, a user needs only to segment multiple background images into several boxes by indicating their vertices as the corresponding points. From these data, the 3D coordinates of the vertices are calculated by using a stereo vision algorithm.

This paper proposes a method to reconstruct the 3D shape model of background regions simply for generating a 3D video, As the progress of computer vision technologies continues, many 3D video systems have been developed that tend to focus on automatic 3D reconstruction to capture target (foreground) objects. On the other hand, the 3D shapes of background regions such as walls and floors are generally made manually using 3D modeling software, i.e., CAD, or as a textured cube that just assumes the room. If we make background models using 3D modeling software, the quality is very high but it costs a lot of time and money. On the other hand, just making a cube is too simple to represent the outer space with sufficient appearance. To solve this problem, we propose a method to reconstruct the background model with simple operations but whose quality remains higher than cubes. In our method, a user needs only to segment multiple background images into several boxes by indicating their vertices as the corresponding points. Prom these data, the 3D coordinates of the vertices are calculated by using a stereo vision algorithm.

収録刊行物

  • 情報処理学会研究報告数理モデル化と問題解決(MPS)

    情報処理学会研究報告数理モデル化と問題解決(MPS) 2008(65(2008-MPS-070)), 1-4, 2008-07-07

    一般社団法人情報処理学会

参考文献:  6件中 1-6件 を表示

各種コード

  • NII論文ID(NAID)
    110006865847
  • NII書誌ID(NCID)
    AN10505667
  • 本文言語コード
    JPN
  • 資料種別
    Technical Report
  • ISSN
    09196072
  • NDL 記事登録ID
    9605363
  • NDL 雑誌分類
    ZM13(科学技術--科学技術一般--データ処理・計算機)
  • NDL 請求記号
    Z14-1121
  • データ提供元
    CJP書誌  NDL  NII-ELS  IPSJ 
ページトップへ