In most railway companies in China, the loss of nuts from rail fasteners is still found by visual inspections performed by maintenance workers; this approach cannot meet the continuing requirements to reduce maintenance costs and improve maintenance efficiency. In this paper, a computer vision system for detecting the loss of nuts from rail fasteners is proposed that is based on kernel two-dimensional principal component -- two-dimensional principal component analysis (K2DPCA-2DPCA) and a support vector machine (SVM). The framework of this information-technology-based system is introduced, and the calculation method to find the positions where nuts are missing is also proposed. The K2DPCA-2DPCA feature extraction method and SVM used for feature classification in the computer vision system are described in detail. Finally, the combination of K2DPCA-2DPCA and SVM is compared with other feature extraction and feature classification methods in an experiment on the detection of the loss of nuts from rail fasteners. In this paper, it is found that the algorithm based on K2DPCA-2DPCA and a SVM is a better method for the computer vision system. It can identify the loss of nuts from a rail fastener with a high recognition rate and reach the maximum recognition rate using only a relatively small number of features. Furthermore, for the same number of training samples, the algorithm based on K2DPCA-2DPCA and SVM is also better than other algorithms for use in the computer vision system.
Computer vision system for detecting the loss of rail fastening nuts based on kernel two-dimensional principal component - two-dimensional principal component analysis and a support vector machine
2016
Article (Journal)
English
Robust principal component analysis for computer vision
IEEE | 2001
|Robust Principal Component Analysis for Computer Vision
British Library Conference Proceedings | 2001
|High Dimensional Model Representation With Principal Component Analysis
Online Contents | 2014
|