Latest revision as of 04:16, 2 February 2021

Abstract

oil pipeline inspection is a dangerous endeavor in the current practice, where a pilot flying in a general aviation class aircraft flies slowly at low altitudes while concurrently looking at the ground for pipeline hazards with the unaided eye; high pilot workload in a dangerous low-speed, low-altitude environment results in an unacceptable number of accidents and loss of life each year. Automation of image acquisition and threat recognition has the potential to reduce pilot workload, improving the safety of the pilots and increasing efficiency. Towards these goals, this paper describes an image classification architecture and algorithm that utilizes several classifiers on different features extracted from the image to automate the threat detection process. The resulting classifier meets the requirement of greater than 80% accuracy in classification. The results will be discussed, and improvements will be proposed for continued research.

Original document

The different versions of the original document can be found in:

• https://zenodo.org/record/1271618 under the license https://creativecommons.org/share-your-work/public-domain/cc0/

• https://zenodo.org/record/1271618/files/article.pdf

• http://xplorestaging.ieee.org/ielx5/5678851/5689839/05689985.pdf?arnumber=5689985,

http://dx.doi.org/10.1109/icsens.2010.5689985 under the license cc0

• http://yadda.icm.edu.pl/yadda/element/bwmeta1.element.ieee-000005689985,

http://www.anefian.com/research/ippolito10_object.pdf,

http://ieeexplore.ieee.org/document/5689985,

https://academic.microsoft.com/#/detail/2106086381