DEFLECTION COMPUTATION OF PIPELINE SURFACE BASED ON 3D POINT CLOUD FROM DEPTH CAMERA

DOI Web Site 2 Citations 2 References
  • INOUE Hiroki
    関西大学理工学研究科環境都市工学専攻都市システム工学分野
  • YASUMURO Yoshihiro
    関西大学環境都市工学部都市システム工学科
  • DAN Hiroshige
    関西大学環境都市工学部都市システム工学科
  • KOBAYASHI Akira
    関西大学環境都市工学部都市システム工学科

Bibliographic Information

Other Title
  • 三次元画像計測による点群データからのパイプラインの変形量の取得

Description

 Numerous developed countries are currently facing problems resulting from the aging of decades-old infra-structures built during periods of rapid economic growth. Additionally, under prevailing severe fiscal circumstances found in many aging societies, it is not realistic to deal with every failing infrastructure facility in every part of a country. This paper focuses on pipelines for water supply, sewerage, and agricultural water use, many of which are buried, which makes investigating their distortion and deterioration difficult without excavation. Under such conditions, maintenance work proceeds slowly and failing buried pipes eventually rupture, which frequently inconvenience citizens and damage property. The authors have been proposing an investigation system that utilizes an RGB-D camera, or a depth-imaging device to examine pipeline interiors. RGB-D cameras can collect not only the RGB-color imagery, but also three-dimensional (3D) depth information from pipe interiors. Taking advantage of the com-pact size of RGB-D cameras, our proposed concept is based on the construction of a self-propelled robot system that will scan pipeline interiors in-situ, thus providing a means of conducting fast and inexpensive investigations that do not require large-scale excavations. This paper describes a novel method that can be used to detect deflection in pipe-shaped structures directly based on depth image capturing with a RGB-D camera. In order to settle intersection planes for deflection examination, the key idea is to find the longitudinal direction from the captured depth images based on the normal vector information, even in the cases that captured 3D point distribution does not form an entire piped-shape. This paper also shows that, based on a prototype implementation using a Kinect camera, the proposed method shows effective performances for actual point cloud data with light computational costs.

Journal

Citations (2)*help

See more

References(2)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top