Font Size: a A A

Study And Realization Of The Train's Dynamic Offset Inspection Based On Machine Vision

Posted on:2009-03-08Degree:MasterType:Thesis
Country:ChinaCandidate:X B ZhengFull Text:PDF
GTID:2178360245982125Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Train will vibrate in all directions when it is running on the railway, and its bodywork will sway to exceed the safeguard limit that result into accident of digression or bump in the extreme condition. In order to ensure a safe condition of the running train and to provide safeguard measurement, it is necessarily to study the technology of the inspection of the train's dynamic offset.When the train is running, the environment is complex and changed greatly, so, it is difficult to inspect the dynamic offset of the train's bodywork in real-time in such condition. Nearly all of its research are in the stage of feasibility argument in native or abroad, still yet to say the appearance of mature production.In this paper, machine vision technology is applied to inspect the train's dynamic offset that is compared to the position of railway. The method of monocular vision and the method of similar binocular stereo vision are provided according to the railway quality separately. Two fast algorithms are suggested separately to satisfy the need of real time, which are the fast fitness of the edges of railway algorithm based on map searching and the max correlation matching algorithm based on alterable gray threshold. In order to weaken the influence of the outside light and segment the foreground effectively, both narrowband LED light and narrowband filter are adopted, moreover, not only the angle of irradiation but also the angle of the shooting of the fast line scan CCD camera is not vertical.The experiments and running results indicate that the inspection system is designed craftily, the algorithm is fast and robust and ideal effect has acquired.
Keywords/Search Tags:machine vision, train's dynamic offset, stereo vision, edge fitness, correlation matching
PDF Full Text Request
Related items