Assessment of a camera pose algorithm using images of brick sewers

Further developments in an ongoing multi-disciplinary research programme concerning the automation of sewer surveys are reported. Previous papers have suggested a theoretical model for use in establishing a frame of reference for closed circuit television (CCTV) camera images of non-man-entry (NME) brick sewers, in order to enable quantitative observations to be automatically acquired employing computer vision techniques. Herein is discussed a simulation testing strategy designed to determine the model's predictive accuracy and thereby assess the corresponding assumptions made. Test results are presented which display robust model characteristics. Further developments are then considered in the context of in-service practice. (C) 2001 Elsevier Science B.V. All rights reserved.