論文

査読有り
2018年

Auto-calibration Method for Active 3D Endoscope System Using Silhouette of Pattern Projector

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
  • Ryo Furukawa
  • ,
  • Masahito Naito
  • ,
  • Daisuke Miyazaki
  • ,
  • Masashi Baba
  • ,
  • Shinsaku Hiura
  • ,
  • Yoji Sanomura
  • ,
  • Shinji Tanaka
  • ,
  • Hiroshi Kawasaki

10749
開始ページ
222
終了ページ
236
記述言語
英語
掲載種別
研究論文(国際会議プロシーディングス)
DOI
10.1007/978-3-319-75786-5_19
出版者・発行元
Springer Verlag

In this paper, we develop an active stereo system for endoscope which requires auto-calibration, because a micro pattern projector is inserted through the instrument channel during an operation and cannot be fixed to the endoscope. For solution, a new auto-calibration technique with full 6-DOF estimation of an active stereo system without any extra devices nor extra pattern projections is proposed. In the technique, the pattern projector itself is simultaneously captured with a target scene by an endoscope camera and the silhouette of the pattern projector is used to conduct 2D-3D matching by using the knowledge of the shape of the projector. In addition, the markers which is included in the projection pattern are extracted and the distances from the closest epipolar lines are calculated as for the cost function. To enhance the robustness of the reconstruction, we also propose a simple high dynamic range (HDR) imaging system for an endoscope by alternating the input power of the pattern projector ON and OFF to blink the pattern so that exposure time will vary with beat frequency, realizing a virtual multi-exposure camera. By applying our auto calibration technique with HDR imaging system, we achieved a robust and accurate reconstruction of tissue in metric 3D under practical operation of the endoscopic system, such as reconstruction of the inside of a real stomach of a pig.

リンク情報
DOI
https://doi.org/10.1007/978-3-319-75786-5_19
ID情報
  • DOI : 10.1007/978-3-319-75786-5_19
  • ISSN : 1611-3349
  • ISSN : 0302-9743
  • SCOPUS ID : 85042495712

エクスポート
BibTeX RIS