Image mosaicing for automated pipe scanning

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

Abstract

Remote visual inspection (RVI) is critical for the inspection of the interior condition of pipelines particularly in the nuclear, oil and gas industries. Conventional RVI equipment produces a video which is analyzed online by a trained inspector employing expert knowledge. Due to the potentially disorientating nature of the footage, this is a time intensive and error prone activity. In this paper a new probe for such visual inspections is presented. The device employs a catadioptric lens coupled with feature based structure from motion to create a 3D model of the interior surface of a pipeline. Reliance upon the availability of image features is mitigated through orientation and distance estimates from an inertial measurement unit and encoder respectively. Such a model affords a global view of the data thus permitting a greater appreciation of the nature and extent of defects. Furthermore, the technique estimates the 3D position and orientation of the probe thus providing information to direct remedial action. Results are presented for both synthetic and real pipe sections. The former enables the accuracy of the generated model to be assessed while the latter demonstrates the efficacy of the technique in a practice.
Original languageEnglish
Title of host publicationE-Book of Abstracts 41st Annual Review of Progress in Quantitative Nondestructive Evaluation Conference
Place of PublicationAmes, Iowa
Publication statusPublished - 1 Oct 2014
Event41st Annual Review of Progress in Quantitative Nondestructive Evaluation Conference - Boise Centre, Boise, United States
Duration: 20 Jul 201425 Jul 2014

Conference

Conference41st Annual Review of Progress in Quantitative Nondestructive Evaluation Conference
Abbreviated titleQNDE
CountryUnited States
CityBoise
Period20/07/1425/07/14

Keywords

  • non destructive evaluation
  • NDE
  • pipe scanning
  • image mosaicing

Fingerprint Dive into the research topics of 'Image mosaicing for automated pipe scanning'. Together they form a unique fingerprint.

Cite this