Research On Security Monitoring Technology Of Subway Based Or Image Mosaic

With the rapid development of urban rail transit,higher requirements are put forward for the safety and emergency response ability of subway.In traditional train monitoring,multiple independent cameras are used to monitor different parts of subway.This monitoring method has problems of low resolution,narrow vision and inconsistent angles of view.When there are emergencies,such as door failure and human stampedes,staff members have to switch back and forth between multiple monitoring screens to determine the cause of the accident so that rescues are delayed.Therefore,the subway security monitoring system is great significant to improve the safety of subway operation and video stitching technology will help achieve this goal.Video stitching technology is based on image mosaic and can stitch multichannel information-related videos together,achieving purposes of extending vision and panoramic monitoring.However,due to the constraints of real time and robustness,traditional video stitching algorithm is unable to be competent for the panoramic monitoring in subway environment.In order to solve these problems,this thesis focuses on image mosaic key technology and proposes the subway safety monitoring system based on stitching technology.The main research work is as follows:1.Based on the particularity of subway environment,the framework of the subway safety monitoring system based on stitching technology was designed to overcome disadvantages of traditional separate video monitoring.As the four components of the system,video collection module,data transmission module,data processing module and subway panoramic video stitching algorithm were designed respectively.Main body of stitching algorithm consisted of image feature detection scheme,image feature matching scheme and video stitching scheme.2.Typical feature description operators such as SUSAN,Moravec,Harris and SIFT were researched.In order to overcome the constraints of the traditional feature description operators in adaptability and robustness,the feature detection scheme based on PCA-SURF feature description operator was designed for the monitoring environment of subway.The scheme achieved dimension reduction with advantages of scale,rotation and illumination invariances and succeeded in improving the computational efficiency.3.Image feature matching was researched.The matching scheme of "two-step" was proposed to overcome problems of time-consuming and low precision in direct matching.Firstly,on basis of researching and absorbing advantages of Laplace response symbol,KD-tree etc.algorithms,initial matching of features was implemented.Then RANSAC extracted features to achieve accurately matching in test environment with complex interference factors.4.Image fusion and video stitching were researched.Video stitching scheme based on the fast index table for subway environment was proposed.The test platform based on the design of the monitoring system was built.According to the platform test results,the subway security monitoring system of this thesis was able to achieve panoramic video monitoring in simulated train environment.Finally,directions of further research were put forward on the basis of summarizing the content of the thesis.