自律型水中ロボットによる鹿児島湾たぎり噴気帯の3次元画像マッピング(第3報)-測位情報と視覚的特徴の併用による画像モザイキング手法-

DOI Web Site

書誌事項

タイトル別名
  • 3D photo mosaicing of Tagiri shallow vent field by an autonomous underwater vehicle (3rd report) - Mosaicing method based on navigation data and visual features -
  • ジリツガタ スイチュウ ロボット ニ ヨル カゴシマワンタギリ フンキタイ ノ 3ジゲン ガゾウ マッピング ダイ3ポウ ソクイ ジョウホウ ト シカクテキ トクチョウ ノ ヘイヨウ ニ ヨル ガゾウ モザイキング シュホウ

この論文をさがす

抄録

Large-area seafloor imaging will bring significant benefits to various fields such as academics, resource survey, marine development, security, and search-and-rescue. The authors have proposed a navigation method of an autonomous underwater vehicle for seafloor imaging, and verified its performance through mapping tubeworm colonies with the area of 3,000 square meters using the AUV Tri-Dog 1 at Tagiri vent field, Kagoshima bay in Japan (Maki et al., 2008, 2009).<br>This paper proposes a post-processing method to build a natural photo mosaic from a number of pictures taken by an underwater platform. The method firstly removes lens distortion, invariances of color and lighting from each image, and then ortho-rectification is performed based on camera pose and seafloor estimated by navigation data. The image alignment is based on both navigation data and visual characteristics, implemented as an expansion of the image based method (Pizarro et al., 2003). Using the two types of information realizes an image alignment that is consistent both globally and locally, as well as making the method applicable to data sets with little visual keys.<br>The method was evaluated using a data set obtained by the AUV Tri-Dog 1 at the vent field in Sep. 2009. A seamless, uniformly illuminated photo mosaic covering the area of around 500 square meters was created from 391 pictures, which covers unique features of the field such as bacteria mats and tubeworm colonies.

収録刊行物

関連プロジェクト

もっと見る

詳細情報 詳細情報について

問題の指摘

ページトップへ