A Novel Approach to Detect and Classify the Defective of Missing Rail Anchors in Real-time

Abstract

In this paper, we have designed a machine visionbased method to identify the defected or missed or grounded rail anchors/fasteners that attach each line with the sleepers. Rail line inspection system is executed manually, especially in third world countries, like Bangladesh. This manual inspection is lengthy, laborious and subjective. To detect and classify the damaged or missed or grounded rail anchors/fasteners in real time based on Shi Tomasi and Harris – Stephen feature detection algorithms with an accuracy of 81.25%. Besides SVM classifier is used to train the Shi Tomasi and Harris – Stephen features which helps to improve the recognition accuracy. To check the robustness of this system, it was tested against on different videos which contain damaged or missed or grounded fasteners in rail line which indicates clearly the high robustness targeted by this system. Keywords—Rail, Anchors, Shi-Tomasi, Harris, SVM

14 Figures and Tables

Cite this paper

@inproceedings{Biswas2017ANA, title={A Novel Approach to Detect and Classify the Defective of Missing Rail Anchors in Real-time}, author={Rubel Biswas and Ahmed Riaz Khan and Samiul Islam and Jia Uddin}, year={2017} }