Optimization Of Railroad Track Flaw Inspection System

Railroad track flaws detection is one of the most important approaches to ensure the safety of railway transportation. In China, railroad track flaws are tested by manual work till now. Therefore, developing an automated system based on image processing and analyses for railroad track flaws detection has great meanings. Compared with manual test, computer automatic test is more accurate and more feasible.In this paper, the requirements of this system and the real images are firstly analyzed, then, based on the previous algorithms and the requirements of real time processing, the algorithms are brought forward as following:1. Railroad track region is segmented from the other regions by use of convolution, differentiation, projection.2. Convoluting the original images with flaws using the spline wavelet, Haar wavelet or square wave, the result of convolution is compared with a threshold to extract the flaws.3. Modifying the edge tracking algorithm of the flaws, using a faster edge tracking algorithm based on image line.Experiments were carried out on the rail images taken from the railroad track, experimental results showed that there were still some problems which are difficult to deal with. For these problems, some attempts and analysis are done in this thesis.