Font Size: a A A

An Abnormal Detection Algorithm Of The EMU Bottom Based On The SURF Features Of The Track-side Image

Posted on:2017-05-21Degree:MasterType:Thesis
Country:ChinaCandidate:D PengFull Text:PDF
GTID:2308330482979439Subject:Electronic Science and Technology
Abstract/Summary:PDF Full Text Request
As one newly popular transportation method,the safety of the EMU has attracted much attention. At present, the main way to detect the EMU abnormal situation is depending on manual work and computer-aided manual work.On the one hand, the structure of the EMU bottom is complex and contains great amount of small parts; on the other hand, traditional way of detection relies on the operator’s technical level and human fatigue degree. As a result, the detection process is hard to avoid slips, even be a threat to the safe operation of the EMU. In order to improve the efficiency and quality of detection, applying computer vision technology to abnormal detection of the bottom of EMU is a smart choice. This paper aims to applying feature extraction and image registration to the track-side image, realizing the automatic detection of abnormal by comparing with the historical standard images. The system designed is supposed to improve the detection efficiency and detection quality.On the basis of existing abnormal detection system, this paper analyses the characteristics of the rail edge image of the EMU bottom. Aiming at the problem of image distortion caused by acquisition process, a method consist of wheel alignment and subsection scaling is been put forward. According to the characteristics of the track image is sensitive to light, the paper adopt the SURF features to the matching process. And added a threshold adaptive step to the match process to improve the registration accuracy and ensure the success rate of registration. After the comparison module, the system analyses the difference images between the breakdown and interference, and designs an algorithm based on image connected domain information, edge information and histogram to judge the abnormal situation.This paper discussed a simulation system based on C++ programming language and OpenCV platform. The simulation experiment of the track-side image of the EMU is carried out which proves that the algorithm can automatically identify the abnormal information at the bottom of the EMU with a reliable efficiency and quality.
Keywords/Search Tags:EMU detection, image processing, SURF feature, abnormal identification
PDF Full Text Request
Related items