Autonomous Rail Track Inspection using Vision Based System

Abstract

This paper proposes a rail track inspection technique using automated video analysis. This system is aimed to replace manual visual checks performed by the railway engineers for track inspection. We suggest a combination of image processing and analysis methods to achieve high performance automated rail track inspection. This paper addresses the problem of finding missing clips and finding blue clips which have been recently replaced in place of damaged clips. The experimental results show high performance in machine vision based inspection on a large sample of real train video