A robust railway fastener missing detection algorithm utilizing image processing

pattern recognition technology for railway digital image acquired by high-speed scanning camera is presented.Aiming at problem of amount of extracted feature is excessive of traditional histograms of oriented gradients（HOG） method

this algorithm firstly preprocesses image to reduce effect of light and noise

use HOG features of ROI area and then reduce dimensions by nuclear-based principal component analysis

through classification training of support vector machine（SVM）

obtain SVM classifier.Compared with traditional HOG+SVM recognition methods

recognition precision and efficiency of this method is greatly improved.