Dissertation > Transportation > Rail transport > Railway Line Project > Railway maintenance and repair > Line detection equipment, test automation

Research on Novelty Detection for High-Speed Railway

Author WangYongLiang
Tutor LuoSiWei;LiQingYong
School Beijing Jiaotong University
Course Computer Science and Technology
Keywords Railway foreign body detection Background modeling Gaussian mixture model Bayesian model Color difference Local Binary Templates
CLC U216.3
Type Master's thesis
Year 2011
Downloads 69
Quotes 0
Download Dissertation

Rail transport has been a major passenger and cargo ways of the country at this stage . In order to further increase the transportation capacity of the railway , China in recent years has been vigorously developing high-speed rail . Railway infrastructure protection is not strong enough to lead to railway accidents occur frequently , and therefore , how to effectively reduce the high-speed rail accident has become a social issue of universal concern . Security measures based on the high-speed rail technology in video surveillance because of its many advantages , starting to get people's attention and attention . The research content is processed by high-speed rail inspections the train shooting the video , automatic detection and analysis of the presence of the security risks of foreign body . The main work of the paper are: 1 suited to the characteristics of the high-speed railway foreign body detection and combined to form a complete solution with Gaussian mixture model and Bayesian model . The characteristics of both the exclusion shadow or remove noise has good performance compared with conventional feature . 2 , improved texture - based modeling , image subregional modeling , makes the detection method is not sensitive to noise , and run faster . 3 , the design and development of a high-speed railway foreign body detection prototype system based on OpenCV . The system implements a variety of detection solutions , have a better performance in the detection precision, recall rate , and real-time .

Related Dissertations
More Dissertations